1008-00005 · Lumos Analytics Inc · Pleadings

ex-parte-tro-application.docx DOCUMENT Litigation

21 / 35 ← → ✕

SUPERIOR COURT OF THE STATE OF CALIFORNIA

COUNTY OF SANTA CLARA

UNLIMITED CIVIL JURISDICTION

C&H Matter No. 1008-00005

CALDERWOOD & HARKNESS LLP Marcus Delgado (SBN 198742), Responsible Partner Catherine Bloom (SBN 214508), Billing Partner Wendy Schultz (SBN 287613), Lead Associate Omar Siddiqui (SBN 312874), Working Associate 555 California Street, 40th Floor San Francisco, California 94104 Telephone: (415) 773-5000 Facsimile: (415) 773-5001

Email: mdelgado@calderwoodharkness.com

Attorneys for Plaintiff LUMOS ANALYTICS INC.

I. NOTICE OF EX PARTE APPLICATION

A. Statement of Application

Plaintiff Lumos Analytics Inc. ("Lumos" or "Plaintiff"), a Delaware corporation with its principal place of business in San Jose, California, hereby applies ex parte for a Temporary Restraining Order ("TRO") and an Order to Show Cause Re: Preliminary Injunction against Defendants NovaSpark Technologies Inc. ("NovaSpark"), Rajiv Anand ("Anand"), Serena Mallick ("Mallick"), and Derek Yun ("Yun") (collectively, "Defendants").

This application is made pursuant to California Code of Civil Procedure sections 527(a) and 527(c), and California Rules of Court, Rule 3.1201 et seq. Lumos respectfully submits that the circumstances set forth herein and in the supporting declarations demonstrate immediate and irreparable injury that will result before the matter can be heard on regular notice, thereby warranting the extraordinary relief of a temporary restraining order issued on an ex parte basis.

This ex parte application is filed concurrently with the Verified Complaint for Trade Secret Misappropriation, Breach of Contract, Breach of Duty of Loyalty, Unfair Business Practices, and Conversion (the "Verified Complaint"), Case No. 23-CV-004871, and is supported by the Memorandum of Points and Authorities incorporated herein, the Declaration of Priya Chandrasekaran, and the Declaration of Rachel Fong.

B. Relief Requested

By this application, Lumos seeks the following emergency relief:

1. Temporary Restraining Order requiring Defendants and each of them, and their officers, directors, agents, employees, representatives, successors, assigns, and all persons acting in concert or participation with them, to:

(1) Immediately cease all use, disclosure, dissemination, and exploitation of Lumos proprietary information and trade secrets, including but not limited to: the LumosEdge ensemble machine-learning model architecture, proprietary feature-engineering libraries, the DataStream ETL pipeline codebase, curated training datasets, and the confidential LumosEdge 3.0 Product Roadmap;

(2) Preserve all electronic devices, storage media, cloud storage accounts, and any and all repositories containing or potentially containing Lumos proprietary information, including but not limited to personal computers, laptops, external hard drives, USB storage devices, smartphones, tablets, cloud accounts (including Hollowell Health IT, Larkspur Technologies, and any other cloud-based storage services), email accounts, NovaSpark servers and development environments, and any other medium in any form in which Lumos proprietary information may be stored;

(3) Refrain from deleting, destroying, altering, overwriting, or concealing any documents, data, files, source code, compiled code, datasets, communications, or electronic information relating to Lumos or its proprietary technology, or relating to any of the Defendants' acquisition, use, or disclosure of Lumos proprietary information;

2. Order to Show Cause why a Preliminary Injunction should not issue on the same terms as the TRO, or on such broader terms as this Court deems appropriate, with a hearing to be set by the Court at the earliest available date.

C. Notice to Defendants

Pursuant to California Rules of Court, Rule 3.1204, Plaintiff's counsel, Calderwood & Harkness LLP, provided notice of this ex parte application to Defendants by email and telephone on January 18, 2023, at approximately 3:00 p.m. Pacific Standard Time. Specifically:

(a) Marcus Delgado of Calderwood & Harkness LLP placed telephone calls and sent email correspondence to Rajiv Anand at his known personal email address (r.anand.personal@vaultmail.com) and personal telephone number on file with Lumos, informing him of the nature of this ex parte application and of the date and time at which it would be presented to the Court.

(b) Marcus Delgado placed telephone calls and sent email correspondence to Serena Mallick at her known personal email address and personal telephone number on file with Lumos, providing identical notice.

(c) Marcus Delgado placed telephone calls and sent email correspondence to Derek Yun at his known personal email address and personal telephone number on file with Lumos, providing identical notice.

(d) Marcus Delgado sent email correspondence and placed a telephone call to NovaSpark Technologies Inc. at its principal address at 88 Bluxome Street, Floor 4, San Francisco, California 94107, directed to the attention of Rajiv Anand in his capacity as a principal of NovaSpark.

As of the filing of this application, no counsel of record has appeared on behalf of any of the Defendants in this action. The Verified Complaint and this ex parte application, together with all supporting papers, have been served on each of the Defendants. Defendants may appear through counsel of their choosing at the hearing on this application or on the Order to Show Cause.

NovaSpark and the individual Defendants have been served with the Verified Complaint and this application by personal service and, where personal service was not immediately achievable, by substituted service in compliance with the California Code of Civil Procedure.

D. Accompanying Papers

This ex parte application is supported by and submitted concurrently with the following papers:

1. Verified Complaint for Trade Secret Misappropriation, Breach of Contract, Breach of Duty of Loyalty, Unfair Business Practices, and Conversion (filed concurrently, Case No. 23-CV-004871) (Exhibit A);

2. Declaration of Priya Chandrasekaran, Chief Executive Officer of Lumos Analytics Inc., with exhibits (Exhibit B);

3. Declaration of Rachel Fong, Vice President of Engineering of Lumos Analytics Inc., with exhibits (Exhibit C);

4. Memorandum of Points and Authorities in Support of Ex Parte Application for Temporary Restraining Order and Order to Show Cause Re: Preliminary Injunction (incorporated herein as Section IV);

5. [Proposed] Temporary Restraining Order (Exhibit D);

6. [Proposed] Order to Show Cause Re: Preliminary Injunction (Exhibit E).

II. INTRODUCTION

This case arises from a calculated and coordinated scheme by three former senior engineers of Lumos Analytics Inc. to steal the company's most valuable and closely guarded trade secrets and use them to launch a direct competitor — NovaSpark Technologies Inc. — virtually overnight.

Lumos Analytics Inc. is a Delaware-incorporated, California-headquartered software-as-a-service ("SaaS") company specializing in artificial-intelligence-driven financial analytics for institutional asset managers. Lumos's principal place of business is located at 4700 Orchard Parkway, Suite 300, San Jose, California 95134. Founded in approximately 2017, Lumos has raised a total of $185 million in venture capital funding through its Series D round, employs approximately 340 people, and generated approximately $67 million in annual recurring revenue ("ARR") in 2022. Lumos's flagship product, LumosEdge, is a proprietary platform that uses ensemble machine-learning models to generate predictive signals for equity and fixed-income portfolio optimization. LumosEdge serves some of the most sophisticated institutional investors in the world and represents the culmination of years of intensive research and development, millions of dollars of capital investment, and the combined expertise of Lumos's engineering and data-science teams.

Three former senior engineers — Defendants Rajiv Anand, Serena Mallick, and Derek Yun — occupied positions of trust and responsibility at Lumos. Anand served as Principal Engineer — ML Infrastructure and was a key architect of the LumosEdge signal-generation engine. Mallick served as Senior Software Engineer — Data Pipelines and was responsible for building and maintaining the proprietary DataStream ETL pipeline system. Yun served as Machine Learning Engineer II with direct access to the company's proprietary training datasets, model hyperparameters, and feature-engineering libraries. Together, these three individuals possessed access to virtually every component of Lumos's core proprietary technology stack.

Each of these individuals executed a Proprietary Information and Inventions Assignment Agreement ("PIIAA") upon commencing employment with Lumos. Each PIIAA contained comprehensive confidentiality and non-disclosure obligations, invention assignment provisions, non-solicitation covenants with a twelve-month post-separation duration, and express covenants not to use or disclose Lumos trade secrets at any time. The confidentiality obligations of the PIIAAs survive termination of employment for a period of three years.

In September and October 2022, Anand, Mallick, and Yun departed Lumos in rapid succession. The timing and sequence of their departures was anything but coincidental. In the days immediately preceding each resignation, each defendant engaged in systematic exfiltration of Lumos proprietary information. Anand transferred approximately 3.2 gigabytes of files — including model architecture specifications and hyperparameter configuration files for the LumosEdge signal-generation engine — to a personal Hollowell Health IT account, and forwarded the confidential 37-page LumosEdge 3.0 Product Roadmap to his personal email. Mallick cloned the entire DataStream ETL pipeline repository, approximately 780 megabytes in size, to a personal external USB drive. Yun downloaded 47 proprietary Jupyter notebooks containing feature-engineering code and model-training scripts to a personal Larkspur Technologies account. None of these transfers was authorized by Lumos, and each was conducted furtively and outside normal business workflows.

Just three days after Anand's resignation, on October 3, 2022, he incorporated NovaSpark Technologies Inc. in Delaware. NovaSpark purports to offer AI-driven alternative-data analytics for hedge funds and proprietary trading desks — a market directly and substantially overlapping with Lumos's core offering. Within days of their own departures, Mallick and Yun joined NovaSpark — Mallick as its Vice President of Engineering and Yun as a founding engineer. In November 2022, NovaSpark secured $12 million in seed funding from Blackpine Capital Partners LP, enabling Defendants to accelerate development of a competing product built upon Lumos's misappropriated technology.

Lumos discovered the exfiltration on December 12, 2022, when its IT security team identified anomalous data-transfer activity in the course of routine security monitoring. On January 9, 2023, Lumos sent cease-and-desist letters to each individual defendant and to NovaSpark, demanding the immediate return of all Lumos proprietary materials and cessation of any use or disclosure of Lumos trade secrets. As of the date of this filing, no substantive response has been received from any defendant.

The harm to Lumos is ongoing and accelerating with each passing day. NovaSpark is actively building a competing product, and the continued use and dissemination of Lumos's trade secrets deepens the injury with every line of code written, every model trained, and every investor presentation delivered. Absent emergency relief from this Court, Lumos's competitive advantage — painstakingly developed over years and at a cost of hundreds of millions of dollars — will be irretrievably lost. Monetary damages cannot restore the secrecy of trade secrets once disclosed, nor can they recapture the competitive position that Lumos stands to lose.

Lumos therefore seeks an ex parte Temporary Restraining Order to preserve the status quo and to prevent Defendants from continuing to exploit Lumos's proprietary information, destroying critical evidence, or further disseminating Lumos's trade secrets to additional third parties. Lumos further requests that the Court issue an Order to Show Cause Re: Preliminary Injunction, to be heard at the earliest available date, so that the Court may consider whether to extend the injunctive relief pending a full adjudication on the merits.

III. STATEMENT OF FACTS

A. Lumos Analytics Inc. and Its Proprietary Technology

1. Company Background

Lumos Analytics Inc. is a corporation organized under the laws of the State of Delaware (Delaware File No. 6518274; Employer Identification Number 82-4713956) with its principal place of business at 4700 Orchard Parkway, Suite 300, San Jose, California 95134. (Chandrasekaran Decl. ¶ 2.) Lumos was founded in approximately 2017 with the mission of applying advanced machine-learning techniques to the financial analytics domain, with a specific focus on generating predictive signals for institutional asset managers. (Id. ¶ 3.)

Lumos's Chief Executive Officer is Priya Chandrasekaran. Its General Counsel is Thomas Brightwell. Its Vice President of Engineering is Rachel Fong. (Chandrasekaran Decl. ¶ 4.) The company employs approximately 340 individuals, the majority of whom are based in its San Jose headquarters, with additional engineering staff distributed across offices in San Francisco and New York. (Id. ¶ 5.)

Since its founding, Lumos has attracted substantial venture capital investment reflecting the market's recognition of the value and promise of its proprietary technology. Lumos has raised a total of approximately $185 million through its Series D round of financing. Notable rounds include a $40 million Series B led by Ridgeline Ventures in 2019 and a $72 million Series D led by Northstar Growth Equity in June 2022. (Chandrasekaran Decl. ¶ 6.) In 2022, Lumos's annual recurring revenue was approximately $67 million, reflecting strong and growing demand for its LumosEdge platform among institutional investors, including hedge funds, pension funds, endowments, and family offices. (Id. ¶ 7.)

Lumos's flagship product is LumosEdge, a cloud-based analytics platform that uses proprietary ensemble machine-learning models to generate predictive signals for equity and fixed-income portfolio optimization. LumosEdge ingests data from dozens of alternative and traditional data sources, processes that data through a proprietary pipeline, engineers thousands of predictive features, and applies sophisticated ML models to generate signals that enable asset managers to make more informed investment decisions. (Fong Decl. ¶ 3.) LumosEdge represents the culmination of years of intensive research and development by a team of talented engineers and data scientists, and it is the core of Lumos's business and competitive position in the market. (Chandrasekaran Decl. ¶ 8.)

2. The Trade Secrets at Issue

At the heart of this action are five categories of proprietary information that constitute trade secrets of Lumos Analytics Inc. under the California Uniform Trade Secrets Act, California Civil Code section 3426.1(d). Each category derives independent economic value, both actual and potential, from not being generally known to the public or to other persons who can obtain economic value from its disclosure or use. Each category is the subject of efforts that are reasonable under the circumstances to maintain its secrecy. (Fong Decl. ¶¶ 5–20; Chandrasekaran Decl. ¶¶ 10–15.)

The five categories are as follows:

(i) LumosEdge Ensemble ML Model Architecture. The LumosEdge signal-generation engine is powered by a proprietary ensemble of machine-learning models. The specific neural-network topology, hyperparameter configurations, model architecture specifications, weighting schemes, and integration logic that comprise this ensemble are the product of years of iterative development, experimentation, and optimization by Lumos's ML engineering team. These specifications define how LumosEdge processes data and generates predictive signals, and they constitute the technical core of the platform. No component of this architecture has been published or disclosed outside of Lumos. (Fong Decl. ¶¶ 6–8.)

(ii) Proprietary Feature-Engineering Libraries. Lumos has developed a library of 2,847 individually engineered features for financial signal generation. Each feature represents a specific mathematical transformation or derivation applied to raw or processed data inputs, designed to capture predictive relationships relevant to asset pricing and portfolio construction. The development of this feature library required years of R&D investment, experimentation, and domain expertise. The specific features, their formulations, and their integration into the LumosEdge pipeline are not publicly known. (Fong Decl. ¶¶ 9–11.)

(iii) DataStream ETL Pipeline Codebase. The DataStream system is Lumos's proprietary extract, transform, and load ("ETL") pipeline. It governs the ingestion, normalization, validation, and routing of data from 43 alternative-data vendors — including satellite imagery providers, web-scraping services, transaction aggregators, and social-media analytics platforms — into the LumosEdge processing environment. The codebase includes data-schema definitions, connector specifications, error-handling logic, and quality-assurance modules. The DataStream system is critical to LumosEdge's ability to ingest diverse data sources at scale and with fidelity. (Fong Decl. ¶¶ 12–14.)

(iv) Curated Training Dataset. Lumos has assembled and curated a training dataset of approximately 14.6 terabytes comprising historical financial and alternative data, labeled and organized for the purpose of training the LumosEdge ML models. The curation process — which includes the selection, cleaning, labeling, augmentation, and organization of data — is itself a proprietary methodology that reflects years of effort and significant financial investment. The dataset is stored on Lumos's secured cloud infrastructure and is accessible only to authorized personnel. (Fong Decl. ¶¶ 15–17.)

(v) Confidential Product Roadmap. Lumos has developed a 37-page internal document titled "LumosEdge 3.0 Product Roadmap" that sets forth the company's planned product development for the period 2023–2025. This document describes planned new features, technical improvements, target markets, competitive positioning strategies, and projected timelines for release. The roadmap was developed by senior leadership and engineering personnel and was classified as strictly confidential. It has never been published or disclosed to any person outside of Lumos other than certain investors subject to confidentiality obligations. (Chandrasekaran Decl. ¶¶ 12–14; Fong Decl. ¶¶ 18–19.)

These five categories of proprietary information represent years of R&D investment, constitute the core competitive advantage of Lumos in the institutional financial analytics market, and are not publicly available or generally known to persons in the relevant industry. (Chandrasekaran Decl. ¶ 15.)

Lumos maintains robust security measures to protect these trade secrets. Access to the LumosEdge codebase, model architecture specifications, feature libraries, DataStream pipelines, and training datasets is restricted to authorized employees through role-based access controls, multi-factor authentication, and encrypted connections. All employees with access to proprietary information are required to execute the PIIAA as a condition of employment. Lumos's IT security infrastructure includes endpoint detection and response tools, data-loss-prevention software, network monitoring, and regular security audits. Physical access to Lumos's offices is controlled through badge-access systems. Sensitive documents are marked as confidential and subject to internal distribution controls. (Fong Decl. ¶ 20; Chandrasekaran Decl. ¶ 16.)

B. The Individual Defendants and Their Access to Trade Secrets

1. Rajiv Anand

Defendant Rajiv Anand, age 37, was employed by Lumos as Principal Engineer — ML Infrastructure from approximately March 2019 through September 30, 2022. (Chandrasekaran Decl. ¶ 18.) As Principal Engineer, Anand was one of the key architects of the LumosEdge signal-generation engine. He was involved in the design, development, testing, and optimization of the ensemble ML model architecture and had deep familiarity with the neural-network topology, hyperparameter configurations, and integration logic underlying the platform. (Id. ¶ 19; Fong Decl. ¶ 22.)

By virtue of his role, Anand had access to Lumos's most sensitive proprietary codebases and training data. Specifically, Anand had access to all five categories of trade secrets identified above: the LumosEdge ensemble ML model architecture, the proprietary feature-engineering libraries, the DataStream ETL pipeline codebase, the curated training dataset, and the confidential LumosEdge 3.0 Product Roadmap. (Fong Decl. ¶ 23.)

Upon commencing his employment with Lumos, Anand executed a Proprietary Information and Inventions Assignment Agreement dated March 11, 2019. The PIIAA contains, among other provisions: (a) an invention assignment clause assigning to Lumos all right, title, and interest in any inventions, developments, or improvements made during the course of employment; (b) a confidentiality and non-disclosure covenant prohibiting the use or disclosure of Lumos proprietary information during and after employment, with the confidentiality obligation surviving termination for three years; (c) a non-solicitation covenant prohibiting solicitation of Lumos employees for a period of twelve months following separation; and (d) a covenant not to use or disclose Lumos trade secrets at any time. A true and correct copy of Anand's executed PIIAA is attached as Exhibit 1 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 20, Ex. 1.)

Anand's residential address is 1482 Mariposa Avenue, Palo Alto, California 94306. (Chandrasekaran Decl. ¶ 21.)

2. Serena Mallick

Defendant Serena Mallick, age 34, was employed by Lumos as Senior Software Engineer — Data Pipelines from approximately June 2020 through October 14, 2022. (Chandrasekaran Decl. ¶ 22.) In this role, Mallick was responsible for building, maintaining, and enhancing the proprietary DataStream ETL pipelines that form the backbone of LumosEdge's data-ingestion system. (Id. ¶ 23; Fong Decl. ¶ 25.)

Mallick had direct access to the DataStream codebase, data-schema definitions, connector specifications, and related technical specifications. She also had access to portions of the broader LumosEdge system as required for integration testing and pipeline validation. (Fong Decl. ¶ 26.)

Upon commencing her employment, Mallick executed a PIIAA dated June 15, 2020, containing the same material provisions as Anand's PIIAA, including invention assignment, confidentiality and non-disclosure obligations surviving three years post-termination, a twelve-month post-separation non-solicitation covenant, and a covenant not to use or disclose Lumos trade secrets at any time. A true and correct copy of Mallick's executed PIIAA is attached as Exhibit 2 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 24, Ex. 2.)

Mallick's residential address is 578 Guerrero Street, Apartment 3B, San Francisco, California 94110. (Chandrasekaran Decl. ¶ 25.)

3. Derek Yun

Defendant Derek Yun, age 31, was employed by Lumos as Machine Learning Engineer II from approximately January 2021 through October 21, 2022. (Chandrasekaran Decl. ¶ 26.) In this role, Yun was responsible for model training, evaluation, and optimization, and had direct access to proprietary training datasets, model hyperparameters, and feature-engineering libraries. (Id. ¶ 27; Fong Decl. ¶ 28.)

Yun's work involved extensive use of proprietary Jupyter notebooks containing feature-engineering code and model-training scripts that embody Lumos's proprietary methodologies for constructing the 2,847 engineered features comprising the company's feature library. (Fong Decl. ¶ 29.)

Upon commencing his employment, Yun executed a PIIAA dated January 11, 2021, containing the same material provisions as the PIIAAs executed by Anand and Mallick. A true and correct copy of Yun's executed PIIAA is attached as Exhibit 3 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 28, Ex. 3.)

Yun's residential address is 2241 Cowper Street, Apartment 12, Palo Alto, California 94301. (Chandrasekaran Decl. ¶ 29.)

C. The Coordinated Departures and Formation of NovaSpark

The sequence of events surrounding the individual Defendants' departures from Lumos and the formation of NovaSpark Technologies Inc. reveals a carefully orchestrated scheme to misappropriate Lumos's proprietary technology and deploy it at a new, directly competing venture. The relevant timeline is as follows:

September 30, 2022: Rajiv Anand resigns from Lumos, effective immediately, without providing the customary two-week notice period and without returning company materials or completing the standard offboarding procedures. Anand's resignation letter was terse and provided no explanation for his departure. (Chandrasekaran Decl. ¶ 30.)

October 3, 2022: Just three days after his resignation from Lumos, Anand incorporates NovaSpark Technologies Inc. in the State of Delaware (Delaware File No. 7294183; Employer Identification Number 88-3291047). NovaSpark establishes its headquarters at 88 Bluxome Street, Floor 4, San Francisco, California 94107. (Chandrasekaran Decl. ¶ 31, Ex. 4.)

NovaSpark's publicly stated business description indicates that it purports to offer "AI-driven alternative-data analytics for hedge funds and proprietary trading desks." This description identifies a market that directly and substantially overlaps with Lumos's core offering. The speed with which NovaSpark was incorporated — within days of Anand's departure — strongly suggests that the planning and preparatory work for the new venture was conducted while Anand was still employed by Lumos. (Chandrasekaran Decl. ¶ 32.)

October 14, 2022: Serena Mallick resigns from Lumos, effective immediately, without providing the customary notice period or completing standard offboarding procedures. (Chandrasekaran Decl. ¶ 33.)

October 17, 2022: Mallick begins employment at NovaSpark as Vice President of Engineering — only three days after her resignation from Lumos. Her Westmark Health IT profile was updated to reflect the NovaSpark role on or about this date. (Chandrasekaran Decl. ¶ 34.)

October 21, 2022: Derek Yun resigns from Lumos, effective immediately, without providing the customary notice period or completing standard offboarding procedures. (Chandrasekaran Decl. ¶ 35.)

October 24, 2022: Yun begins employment at NovaSpark as a founding engineer — only three days after his resignation from Lumos. (Chandrasekaran Decl. ¶ 36.)

November 15, 2022: NovaSpark closes a $12 million seed financing round from Blackpine Capital Partners LP (Managing Partner: Franklin Osei), a venture capital firm headquartered at 101 Mission Street, 38th Floor, San Francisco, California 94105. The closing of a $12 million seed round within approximately six weeks of NovaSpark's incorporation is unusual and suggests that NovaSpark was able to demonstrate a level of product development and technological sophistication that would ordinarily require many months, if not years, of work — a circumstance that is consistent with and strongly suggests the use of Lumos's misappropriated technology. (Chandrasekaran Decl. ¶ 37.)

The rapid sequence of departures, the immediate formation of a direct competitor, and the near-instantaneous hiring of all three departing engineers at the new entity are not coincidences. Taken together, these facts strongly support the inference that the individual Defendants engaged in a coordinated, premeditated scheme to misappropriate Lumos's trade secrets and use them to build NovaSpark. (Chandrasekaran Decl. ¶ 38.)

D. Evidence of Data Exfiltration

On December 12, 2022, Lumos's IT security team, in the course of routine security monitoring and a targeted post-departure audit of the individual Defendants' system activity, identified anomalous data-exfiltration activity dating to the final weeks of Anand's, Mallick's, and Yun's employment at Lumos. (Fong Decl. ¶ 31.) The internal investigation, conducted under the direction of VP of Engineering Rachel Fong and Lumos's Director of Information Security, Jonathan Mehta, revealed the following specific acts of misappropriation by each individual Defendant:

1. Rajiv Anand

On September 26, 2022 — four days before his resignation — Anand transferred approximately 3.2 gigabytes of files from Lumos's secure internal repositories to a personal Hollowell Health IT account. The transferred files included model architecture specifications, hyperparameter configuration files, and related documentation for the LumosEdge signal-generation engine. The transfer was identified through Lumos's data-loss-prevention ("DLP") system and confirmed through analysis of cloud-sync logs, endpoint monitoring data, and network egress records. (Fong Decl. ¶ 33, Ex. A.)

On September 28, 2022 — two days before his resignation — Anand accessed the confidential 37-page document titled "LumosEdge 3.0 Product Roadmap" from Lumos's internal document-management system and forwarded it as a PDF attachment to his personal email address (r.anand.personal@vaultmail.com). The email was captured by Lumos's email monitoring system, and the access and forwarding event is reflected in the system audit logs. (Fong Decl. ¶ 34, Ex. B.)

These transfers occurred outside Anand's normal business workflow. There was no legitimate business reason for Anand to transfer model architecture files or the product roadmap to personal, non-Lumos accounts. The transfers were not authorized by Lumos management, and no authorization was sought. (Fong Decl. ¶ 35.)

2. Serena Mallick

On October 10, 2022 — four days before her resignation — Mallick connected a personal external USB drive, identified by serial number WD-4837291, to her Lumos-issued workstation. She then cloned the entire DataStream ETL pipeline repository, comprising approximately 780 megabytes of proprietary source code, data-schema definitions, connector specifications, configuration files, and related documentation. The connection of the external device and the file-copy operation were recorded by Lumos's endpoint detection and response ("EDR") system. (Fong Decl. ¶ 37, Ex. C.)

The DataStream repository contains the full codebase and data-schema definitions for Lumos's proprietary data-ingestion system covering 43 alternative-data vendors. The repository is maintained in Lumos's private Git instance and is accessible only to authorized engineering personnel through authenticated, encrypted connections. (Fong Decl. ¶ 38.)

The cloning of the entire repository to a personal external device was not authorized by Lumos, was outside the scope of Mallick's normal duties, and was contrary to Lumos's information-security policies, which expressly prohibit the copying of proprietary source code to personal or unauthorized devices. (Fong Decl. ¶ 39.)

3. Derek Yun

On October 18, 2022 — three days before his resignation — Yun downloaded 47 proprietary Jupyter notebooks from Lumos's internal development environment to a personal Larkspur Technologies account. The downloaded notebooks contained feature-engineering code and model-training scripts that embody Lumos's proprietary methodologies for constructing the 2,847 engineered features that form the core of its feature-engineering library. The download was identified through cloud-sync agent logs on Yun's Lumos-issued laptop and confirmed through analysis of Larkspur Technologies API activity captured by the DLP system. (Fong Decl. ¶ 41, Ex. D.)

These Jupyter notebooks represent the practical implementation of Lumos's most closely guarded feature-engineering techniques. Their contents would enable a competitor to replicate or closely approximate the feature-engineering capabilities of LumosEdge without independently investing the years of R&D effort required to develop them. (Fong Decl. ¶ 42.)

The download was not authorized by Lumos. There was no legitimate business reason for Yun to transfer these notebooks to a personal Larkspur Technologies account, and no authorization was sought. (Fong Decl. ¶ 43.)

E. Failure to Return Company Materials

Under the terms of their respective PIIAAs, each individual Defendant was obligated to return all company materials, documents, data, equipment, and proprietary information upon the termination of their employment with Lumos. Section 5 of each PIIAA states, in pertinent part:

"Upon termination of my employment with the Company, I will promptly deliver to the Company, and will not keep in my possession, recreate, or deliver to anyone else, any and all devices, records, data, notes, reports, proposals, lists, correspondence, specifications, drawings, blueprints, sketches, laboratory notebooks, materials, flow charts, equipment, other documents or property, or reproductions of any of the aforementioned items developed by me pursuant to my employment with the Company or otherwise belonging to the Company, its successors, or assigns."

(Chandrasekaran Decl. ¶ 39, Exs. 1–3.)

Despite this clear contractual obligation, none of the three individual Defendants returned company materials upon their departure from Lumos. None of the three confirmed the deletion of Lumos proprietary information from their personal devices or accounts. Anand returned his Lumos-issued laptop in a perfunctory manner but did not return or account for any of the materials he had transferred to his personal Hollowell Health IT or personal email. Mallick returned her Lumos-issued laptop but did not return the external USB drive (serial number WD-4837291) containing the cloned DataStream repository or account for the data on it. Yun returned his Lumos-issued laptop but did not confirm deletion of the Jupyter notebooks from his personal Larkspur Technologies account. (Chandrasekaran Decl. ¶¶ 40–42; Fong Decl. ¶ 44.)

On January 9, 2023, Lumos, through its General Counsel Thomas Brightwell, sent cease-and-desist letters to each of the individual Defendants and to NovaSpark Technologies Inc. by certified mail and email. The letters demanded: (i) the immediate return of all Lumos proprietary materials in Defendants' possession, custody, or control; (ii) the immediate cessation of all use and disclosure of Lumos trade secrets and proprietary information; (iii) confirmation that all copies of Lumos proprietary information had been permanently deleted from all personal devices, accounts, and storage media; and (iv) a written response within seven days. (Chandrasekaran Decl. ¶ 43, Ex. 5.)

As of the date of this filing — January 19, 2023 — no substantive response has been received from any Defendant to the cease-and-desist letters. The silence of the Defendants in the face of these demands heightens the urgency of this application and the risk that Lumos's proprietary information is being used, disseminated, or destroyed. (Chandrasekaran Decl. ¶ 44.)

F. Retention of Counsel and Decision to Seek Emergency Relief

On December 15, 2022, following the discovery of the data-exfiltration activity, Lumos retained Calderwood & Harkness LLP to investigate the circumstances of the individual Defendants' departures and the formation of NovaSpark, and to evaluate and pursue all legal remedies available to Lumos. (Chandrasekaran Decl. ¶ 45.)

Based on its investigation — including review of the IT security findings, analysis of the PIIAAs, examination of publicly available information regarding NovaSpark, and consultations with Lumos's CEO and VP of Engineering — Calderwood & Harkness determined that emergency judicial relief is necessary and warranted. The ongoing nature of the misappropriation, the accelerating development of NovaSpark's competing product, the Defendants' failure to respond to the cease-and-desist letters, and the risk of further dissemination and destruction of evidence all counsel in favor of seeking a TRO on an ex parte basis. Every day of delay increases the irreparable harm to Lumos and diminishes the effectiveness of any eventual judicial remedy. (Chandrasekaran Decl. ¶ 46.)

IV. MEMORANDUM OF POINTS AND AUTHORITIES

A. Legal Standard for Ex Parte Temporary Restraining Order

California Code of Civil Procedure section 527(a) authorizes a court to grant a temporary restraining order upon a showing that "great or irreparable injury" will result to the applicant before the matter can be heard on notice. Cal. Code Civ. Proc. § 527(a). The purpose of a temporary restraining order is to preserve the status quo pending a hearing on a preliminary injunction. Oiye v. Fox (2012) 211 Cal.App.4th 1036, 1047.

California Rules of Court, Rule 3.1201 et seq., governs the procedural requirements for ex parte applications, including requirements for notice to the opposing party and the content of the application. The applicant must make an affirmative factual showing of the necessity for the temporary restraining order by a declaration containing specific facts showing that immediate and irreparable injury, loss, or damage will result if the relief is not granted before the opposing party or its attorney can be heard in opposition. Cal. Rules of Court, Rule 3.1202(c).

To obtain a temporary restraining order, the applicant must demonstrate: (1) a likelihood of prevailing on the merits at trial; (2) that the applicant will suffer irreparable harm in the absence of injunctive relief; (3) that the balance of hardships tips in the applicant's favor; and (4) that the threat of immediate and irreparable injury justifies ex parte relief without full adversarial hearing. White v. Davis (2003) 30 Cal.4th 528, 554; ReadyLink Healthcare v. Cotton (2005) 126 Cal.App.4th 1006, 1016.

In cases involving trade secret misappropriation, the California Uniform Trade Secrets Act ("CUTSA"), codified at California Civil Code sections 3426–3426.11, provides specific statutory authorization for injunctive relief. Section 3426.2(a) provides that "[a]ctual or threatened misappropriation may be enjoined." Cal. Civ. Code § 3426.2(a). This provision reflects the legislative determination that trade secret misappropriation is a category of harm particularly amenable to, and often requiring, injunctive relief. Courts have recognized that trade secret cases present "the classic case for the grant of an injunction" because "[t]he very nature of the secret . . . constitutes an interest that cannot be properly protected by an award of monetary damages." Readylink Healthcare, 126 Cal.App.4th at 1020 (citing DVD Copy Control Ass'n, Inc. v. Bunner (2003) 31 Cal.4th 864, 878).

As demonstrated below, Lumos satisfies each of these requirements.

B. Lumos Is Likely to Prevail on the Merits

1. First Cause of Action: Trade Secret Misappropriation (Cal. Civ. Code §§ 3426–3426.11)

Lumos is likely to prevail on its claim for trade secret misappropriation under the California Uniform Trade Secrets Act. The elements of a claim for trade secret misappropriation under CUTSA are: (a) the existence of a trade secret; and (b) misappropriation of that trade secret by the defendant. Cal. Civ. Code § 3426.1; Sargent Fletcher, Inc. v. Able Corp. (2003) 110 Cal.App.4th 1658, 1665.

Existence of Trade Secrets. A "trade secret" is defined as information that "[d]erives independent economic value, actual or potential, from not being generally known to the public or to other persons who can obtain economic value from its disclosure or use" and "[i]s the subject of efforts that are reasonable under the circumstances to maintain its secrecy." Cal. Civ. Code § 3426.1(d).

The five categories of Lumos proprietary information identified in Section III.A.2 above satisfy both prongs of this definition. First, each category derives independent economic value from not being generally known: the ML model architecture defines the core analytic capability of LumosEdge; the feature-engineering libraries represent thousands of individually developed predictive features; the DataStream pipeline codebase enables the ingestion of data from 43 vendors; the training dataset reflects years of curation; and the product roadmap reveals Lumos's strategic direction for the next three years. (Fong Decl. ¶¶ 5–20; Chandrasekaran Decl. ¶¶ 10–15.) These materials are not publicly available, have not been disclosed outside of Lumos (except under confidentiality obligations to investors), and would confer significant competitive advantage upon any competitor that obtained them. (Chandrasekaran Decl. ¶ 15.)

Second, Lumos has taken reasonable steps to maintain the secrecy of this information, including role-based access controls, multi-factor authentication, encrypted connections, endpoint detection and response systems, data-loss prevention software, physical access controls, mandatory PIIAAs for all employees with access to proprietary information, and internal policies restricting the distribution of confidential documents. (Fong Decl. ¶ 20; Chandrasekaran Decl. ¶ 16.) These measures are more than adequate to satisfy the "reasonable efforts" requirement under California law. See FLIR Systems, Inc. v. Parrish (2009) 174 Cal.App.4th 1270, 1279 (reasonable measures include confidentiality agreements, restricted access, and security protocols).

Misappropriation. "Misappropriation" includes, among other things, the acquisition of a trade secret by a person who knows or has reason to know that the trade secret was acquired by improper means, or the disclosure or use of a trade secret without express or implied consent by a person who used improper means to acquire knowledge of the trade secret, or who knew or had reason to know that knowledge of the trade secret was acquired by improper means. Cal. Civ. Code § 3426.1(b).

Here, the evidence establishes that each individual Defendant acquired Lumos trade secrets through improper means — namely, the unauthorized copying and transfer of proprietary information to personal accounts and devices in the days immediately preceding their resignations. (Fong Decl. ¶¶ 33–43.) These transfers were conducted furtively, outside the scope of the Defendants' normal duties, and in violation of Lumos's information-security policies. The subsequent use of these trade secrets in connection with NovaSpark — a direct competitor incorporated within days of Anand's departure — constitutes ongoing misappropriation through disclosure and use.

NovaSpark is itself liable for misappropriation to the extent it acquired Lumos trade secrets through the individual Defendants with knowledge, or reason to know, that the information was obtained by improper means. Given that NovaSpark was founded by Anand and staffed by Mallick and Yun — the very individuals who exfiltrated the trade secrets — NovaSpark's knowledge is imputed as a matter of law. See Ajaxo Inc. v. ETrade Grp., Inc.* (2005) 135 Cal.App.4th 21, 66.

The documentary evidence — including IT security logs, file-transfer records, email forwarding records, endpoint monitoring data, and cloud-sync logs — provides strong evidentiary support for Lumos's misappropriation claim. (Fong Decl. ¶¶ 31–43, Exs. A–D.)

2. Second Cause of Action: Breach of Contract (PIIAA)

Lumos is also likely to prevail on its claim for breach of the PIIAAs executed by each individual Defendant. Each PIIAA constitutes a valid and enforceable contract between the individual Defendant and Lumos. Each PIIAA contains express obligations including: (a) a duty to maintain the confidentiality of Lumos proprietary information and not to use or disclose it during or after employment; (b) a duty to assign to Lumos all inventions made during the course of employment; (c) a duty to return all company materials upon termination; and (d) a non-solicitation covenant of twelve months' duration. (Chandrasekaran Decl. ¶¶ 20, 24, 28, Exs. 1–3.)

Each individual Defendant breached the PIIAA in multiple respects. First, each Defendant copied and removed Lumos proprietary information — model architecture files, pipeline code, Jupyter notebooks, and the product roadmap — in violation of the confidentiality and non-disclosure provisions. Second, each Defendant failed to return all company materials upon termination, as required by Section 5 of the PIIAA. Third, each Defendant has used and/or disclosed Lumos trade secrets in connection with NovaSpark, in violation of the covenant not to use or disclose Lumos trade secrets at any time.

The existence of the contracts, the specific terms breached, and the specific acts of breach are all supported by documentary evidence. (Chandrasekaran Decl. ¶¶ 39–44; Fong Decl. ¶¶ 31–44.)

3. Third Cause of Action: Breach of Duty of Loyalty

Under California law, an employee owes a duty of loyalty to his or her employer during the period of employment. This duty prohibits an employee from acting in competition with the employer, diverting business opportunities, or misappropriating the employer's property for personal benefit while the employment relationship subsists. Huong Que, Inc. v. Luu (2007) 150 Cal.App.4th 400, 410; Stokes v. Dole Nut Co. (1995) 41 Cal.App.4th 285, 295.

While still employed by Lumos, each individual Defendant engaged in acts of data exfiltration in preparation for competing with Lumos through NovaSpark. Anand's transfers on September 26 and 28, 2022, occurred during his employment at Lumos. Mallick's cloning of the DataStream repository on October 10, 2022, occurred during her employment at Lumos. Yun's download of Jupyter notebooks on October 18, 2022, occurred during his employment at Lumos. (Fong Decl. ¶¶ 33–43.)

These acts — taking proprietary information for the purpose of launching or supporting a competing venture — constitute clear breaches of the common-law duty of loyalty owed by each Defendant to Lumos during the period of employment. See Fowler v. Varian Assocs., Inc. (1987) 196 Cal.App.3d 34, 41 (employees may not misappropriate employer property for use in a competing business while still employed).

4. Fourth Cause of Action: Unfair Business Practices (Cal. Bus. & Prof. Code § 17200)

California Business and Professions Code section 17200 prohibits "any unlawful, unfair, or fraudulent business act or practice." The statute is broadly construed to encompass a wide range of unfair competitive conduct. Cel-Tech Commc'ns, Inc. v. Los Angeles Cellular Tel. Co. (1999) 20 Cal.4th 163, 180.

The Defendants' misappropriation of Lumos's trade secrets, breach of their contractual obligations under the PIIAAs, and breach of their duty of loyalty constitute unlawful business practices within the meaning of section 17200, because these acts violate the California Uniform Trade Secrets Act, applicable contract law, and the common-law duty of loyalty. See Glenrock Corporation (2003) 29 Cal.4th 1134, 1143 (section 17200 "borrows" violations of other laws and treats them as unlawful practices).

Moreover, NovaSpark's use of Lumos's proprietary technology to develop a competing product and gain an unfair competitive advantage constitutes an independently actionable unfair business practice. A business practice is "unfair" under section 17200 if it offends an established public policy or is immoral, unethical, oppressive, unscrupulous, or substantially injurious to consumers or competitors. Cel-Tech, 20 Cal.4th at 186–187. Building a competing business on stolen trade secrets satisfies this standard.

5. Fifth Cause of Action: Conversion

Conversion is "the wrongful exercise of dominion over the property of another." Lee v. Hanley (2015) 61 Cal.4th 1225, 1240. The elements of conversion are: (1) ownership or right to possession of the property; (2) wrongful disposition of the property right; and (3) damages. Id.

California courts have recognized that trade secrets and proprietary information constitute property interests that can be the subject of a conversion claim. See Fremont Indem. Co. v. Fremont Gen. Corp. (2007) 148 Cal.App.4th 97, 125 (conversion can apply to intangible property that is merged into a document or other tangible thing). The Defendants converted Lumos's proprietary information — including tangible embodiments thereof in the form of electronic files, code, and documents — by copying, taking, retaining, and using that information without authorization and contrary to Lumos's property rights. Each Defendant exercised dominion and control over Lumos's property in a manner inconsistent with Lumos's rights of ownership.

C. Lumos Will Suffer Irreparable Harm Absent Injunctive Relief

Trade secret misappropriation, by its very nature, causes irreparable harm. Once a trade secret is disclosed or used by an unauthorized person, the secrecy that gives it value is destroyed and cannot be restored. FLIR Systems, Inc. v. Parrish (2009) 174 Cal.App.4th 1270, 1278 ("[T]he disclosure of a trade secret in violation of a confidential relationship can constitute irreparable injury."); Readylink Healthcare, Inc. v. Cotton (2005) 126 Cal.App.4th 1006, 1024 (finding irreparable harm where trade secrets were being used by a competitor). California Civil Code section 3426.2 itself reflects the legislature's recognition that injunctive relief is the appropriate remedy for actual or threatened misappropriation, precisely because monetary relief is often inadequate. Cal. Civ. Code § 3426.2(a).

Here, the irreparable harm to Lumos is neither speculative nor remote — it is ongoing and accelerating. NovaSpark is actively developing a competing product using Lumos's misappropriated trade secrets. Every day that passes without judicial intervention, the Defendants continue to integrate Lumos's proprietary technology — its model architectures, feature libraries, pipeline code, and strategic product plans — into NovaSpark's systems. The deeper that integration proceeds, the more difficult it becomes to disentangle and the more completely Lumos's competitive advantage is destroyed.

The risk of further dissemination compounds the harm. NovaSpark is not a solo operation; it has received $12 million in funding from Blackpine Capital Partners LP and presumably has or will hire additional employees. Each additional person who gains access to Lumos's trade secrets widens the circle of dissemination and diminishes the secrecy — and therefore the value — of the information. Investor presentations, recruiting materials, and customer demonstrations may further expose Lumos's proprietary technology to parties beyond the Defendants' immediate control.

Monetary damages are inadequate to redress the harm for at least three reasons. First, the value of trade secrets is inherently difficult to quantify once the secrecy has been breached; the harm is not merely the cost of developing the information but the permanent loss of the competitive advantage it confers. (Chandrasekaran Decl. ¶ 47.) Second, Lumos's competitive position — built over years of effort and hundreds of millions of dollars in investment — cannot be recaptured through a damages award after the fact; a competitor that reaches the market with a product built on Lumos's technology will permanently alter the competitive landscape. (Id. ¶ 48.) Third, NovaSpark's accelerated time-to-market, enabled by the misappropriated technology, causes market damage that compounds over time as NovaSpark acquires customers, builds relationships, and establishes brand recognition at Lumos's expense. (Id. ¶ 49.)

Accordingly, Lumos will suffer immediate and irreparable harm absent the issuance of a temporary restraining order.

D. The Balance of Hardships Favors Lumos

The balance of hardships tips decisively in favor of Lumos.

Harm to Lumos absent relief: Lumos stands to lose its core competitive advantage — proprietary technology representing years of research and development and hundreds of millions of dollars of capital investment. LumosEdge is the foundation of Lumos's business and the basis on which it has attracted customers, investors, and employees. If Defendants are permitted to continue using Lumos's trade secrets to build NovaSpark, Lumos's competitive position in the institutional asset management analytics market will be permanently and irreparably impaired. (Chandrasekaran Decl. ¶¶ 47–49.)

Burden on Defendants if relief is granted: The proposed TRO imposes only the requirement that Defendants refrain from using property that does not belong to them and that they preserve evidence relevant to this litigation. The TRO does not prevent NovaSpark from operating as a business, does not prevent the individual Defendants from pursuing their careers, and does not prohibit NovaSpark from competing with Lumos using independently developed technology or publicly available information. The TRO merely requires Defendants to cease exploiting Lumos's stolen trade secrets — an obligation they already owe under their PIIAAs and under the law. Defendants can articulate no legitimate interest in continuing to use property that was acquired through wrongful conduct. See Bramwell Partners LLC v. Hamidi (2003) 30 Cal.4th 1342, 1352 (no cognizable interest in continuing an unlawful act).

The hardship to Defendants of complying with a narrowly tailored restraining order is minimal compared to the hardship to Lumos of losing its trade secrets permanently. The equities overwhelmingly favor Lumos.

E. The Circumstances Justify Ex Parte Relief

Ex parte relief is warranted when there is an immediate threat that the opposing party will destroy evidence or cause further irreparable harm if given advance notice of the application. Cal. Rules of Court, Rule 3.1202(c); see also O'Connell v. Superior Court (2006) 141 Cal.App.4th 1452, 1463.

Here, there is a concrete and immediate threat that Defendants will destroy evidence or further disseminate Lumos's trade secrets if they are given advance notice of the specific terms of the relief sought. The Defendants have already demonstrated a willingness to misappropriate confidential information through covert means — transferring files to personal accounts, cloning repositories to external drives, and forwarding documents to personal email addresses. Individuals who have shown this disregard for their obligations cannot be trusted to preserve the very evidence that proves their wrongdoing.

Moreover, the cease-and-desist letters sent by Lumos on January 9, 2023, demanded the return of materials and cessation of use. The Defendants' failure to respond — ten days of silence — heightens the risk of spoliation. A party that ignores a cease-and-desist demand is more likely, not less, to delete or conceal evidence when it learns that litigation is imminent and that specific forensic inquiries will be directed at its devices and accounts.

Time is of the essence. NovaSpark is actively building its product. The longer Defendants are permitted to use Lumos's trade secrets, the more deeply those secrets become embedded in NovaSpark's systems, the more broadly they are disseminated, and the more difficult it will be to fashion effective relief. Delay in obtaining a TRO will only deepen the harm to Lumos and undermine the Court's ability to preserve the status quo. The urgency of the circumstances fully justifies the ex parte procedure.

F. The Proposed TRO Is Narrowly Tailored

The proposed TRO is narrowly tailored to address the specific harm at issue. The injunctive provisions address only Lumos's proprietary information and trade secrets. They do not prevent Defendants from conducting lawful business activities, pursuing employment in any field, or developing and marketing technology that is independently created or derived from publicly available sources.

The preservation requirements are similarly limited in scope. They require Defendants to maintain the status quo with respect to devices and accounts that may contain Lumos proprietary information. Preservation is necessary both to maintain the evidentiary record for this litigation and to enable forensic examination of Defendants' devices and accounts to determine the full scope of the misappropriation. Without a preservation order, there is a significant risk that critical evidence will be lost — whether through intentional deletion or through the operation of automatic data-retention and destruction policies.

The proposed TRO does not impose any bond requirement beyond the statutory minimum, and Lumos is prepared to post a bond in such amount as the Court may determine is appropriate.

In sum, the proposed TRO is the minimum relief necessary to preserve the status quo, protect Lumos's trade secrets, and maintain the integrity of the evidentiary record pending a full hearing on the preliminary injunction.

V. DESCRIPTION OF PROPOSED TEMPORARY RESTRAINING ORDER

Lumos respectfully submits the proposed Temporary Restraining Order attached hereto as Exhibit D. The key provisions of the proposed order are as follows:

A. Injunctive Provisions

Upon issuance of the TRO, Defendants — and each of them, and their officers, directors, agents, employees, representatives, successors, assigns, and all persons acting in concert or participation with them who receive actual notice of the order — shall immediately cease all use, disclosure, dissemination, distribution, publication, or exploitation of any and all Lumos Analytics Inc. trade secrets and proprietary information, including but not limited to:

(i) The LumosEdge ensemble ML model architecture, including all neural-network topology specifications, hyperparameter configurations, model architecture documents, weighting schemes, and integration logic;

(ii) The proprietary feature-engineering libraries, including all 2,847 individually engineered features, their formulations, and their integration specifications;

(iii) The DataStream ETL pipeline codebase, including all source code, data-schema definitions, connector specifications, configuration files, error-handling logic, and quality-assurance modules;

(iv) The curated training datasets, consisting of approximately 14.6 terabytes of historical financial and alternative data assembled and curated by Lumos for ML model training;

(v) The LumosEdge 3.0 Product Roadmap and all related confidential product development materials, including any excerpts, summaries, or derivative documents.

B. Preservation Provisions

Defendants shall immediately preserve and maintain in their current state all electronic devices, storage media, servers, cloud storage accounts, email accounts, and any and all repositories that contain or may contain Lumos proprietary information. Without limiting the generality of the foregoing, the specific devices and accounts to be preserved include but are not limited to:

(a) Rajiv Anand's personal Hollowell Health IT account (the account to which approximately 3.2 GB of files were transferred on September 26, 2022) and personal email account (r.anand.personal@vaultmail.com);

(b) Serena Mallick's personal external USB drive, serial number WD-4837291, and any copies, backups, or images of the data contained thereon;

(c) Derek Yun's personal Larkspur Technologies account (the account to which 47 Jupyter notebooks were downloaded on October 18, 2022);

(d) All NovaSpark Technologies Inc. servers, cloud infrastructure, code repositories (including but not limited to Git repositories), development environments, staging environments, production environments, and associated backup systems;

(e) All personal laptops, desktop computers, smartphones, tablets, external hard drives, USB storage devices, and any other el

SUPERIOR COURT OF THE STATE OF CALIFORNIA
COUNTY OF SANTA CLARA
UNLIMITED CIVIL JURISDICTION
C&H Matter No. 1008-00005
CALDERWOOD & HARKNESS LLP Marcus Delgado (SBN 198742), Responsible Partner Catherine Bloom (SBN 214508), Billing Partner Wendy Schultz (SBN 287613), Lead Associate Omar Siddiqui (SBN 312874), Working Associate 555 California Street, 40th Floor San Francisco, California 94104 Telephone: (415) 773-5000 Facsimile: (415) 773-5001
Email: mdelgado@calderwoodharkness.com
Attorneys for Plaintiff LUMOS ANALYTICS INC.
TABLE OF CONTENTS
I. NOTICE OF EX PARTE APPLICATION
A. Statement of Application
Plaintiff Lumos Analytics Inc. ("Lumos" or "Plaintiff"), a Delaware corporation with its principal place of business in San Jose, California, hereby applies ex parte for a Temporary Restraining Order ("TRO") and an Order to Show Cause Re: Preliminary Injunction against Defendants NovaSpark Technologies Inc. ("NovaSpark"), Rajiv Anand ("Anand"), Serena Mallick ("Mallick"), and Derek Yun ("Yun") (collectively, "Defendants").
This application is made pursuant to California Code of Civil Procedure sections 527(a) and 527(c), and California Rules of Court, Rule 3.1201 et seq. Lumos respectfully submits that the circumstances set forth herein and in the supporting declarations demonstrate immediate and irreparable injury that will result before the matter can be heard on regular notice, thereby warranting the extraordinary relief of a temporary restraining order issued on an ex parte basis.
This ex parte application is filed concurrently with the Verified Complaint for Trade Secret Misappropriation, Breach of Contract, Breach of Duty of Loyalty, Unfair Business Practices, and Conversion (the "Verified Complaint"), Case No. 23-CV-004871, and is supported by the Memorandum of Points and Authorities incorporated herein, the Declaration of Priya Chandrasekaran, and the Declaration of Rachel Fong.
B. Relief Requested
By this application, Lumos seeks the following emergency relief:
1. Temporary Restraining Order requiring Defendants and each of them, and their officers, directors, agents, employees, representatives, successors, assigns, and all persons acting in concert or participation with them, to:
(1) Immediately cease all use, disclosure, dissemination, and exploitation of Lumos proprietary information and trade secrets, including but not limited to: the LumosEdge ensemble machine-learning model architecture, proprietary feature-engineering libraries, the DataStream ETL pipeline codebase, curated training datasets, and the confidential LumosEdge 3.0 Product Roadmap;
(2) Preserve all electronic devices, storage media, cloud storage accounts, and any and all repositories containing or potentially containing Lumos proprietary information, including but not limited to personal computers, laptops, external hard drives, USB storage devices, smartphones, tablets, cloud accounts (including Hollowell Health IT, Larkspur Technologies, and any other cloud-based storage services), email accounts, NovaSpark servers and development environments, and any other medium in any form in which Lumos proprietary information may be stored;
(3) Refrain from deleting, destroying, altering, overwriting, or concealing any documents, data, files, source code, compiled code, datasets, communications, or electronic information relating to Lumos or its proprietary technology, or relating to any of the Defendants' acquisition, use, or disclosure of Lumos proprietary information;
2. Order to Show Cause why a Preliminary Injunction should not issue on the same terms as the TRO, or on such broader terms as this Court deems appropriate, with a hearing to be set by the Court at the earliest available date.
C. Notice to Defendants
Pursuant to California Rules of Court, Rule 3.1204, Plaintiff's counsel, Calderwood & Harkness LLP, provided notice of this ex parte application to Defendants by email and telephone on January 18, 2023, at approximately 3:00 p.m. Pacific Standard Time. Specifically:
(a) Marcus Delgado of Calderwood & Harkness LLP placed telephone calls and sent email correspondence to Rajiv Anand at his known personal email address (r.anand.personal@vaultmail.com) and personal telephone number on file with Lumos, informing him of the nature of this ex parte application and of the date and time at which it would be presented to the Court.
(b) Marcus Delgado placed telephone calls and sent email correspondence to Serena Mallick at her known personal email address and personal telephone number on file with Lumos, providing identical notice.
(c) Marcus Delgado placed telephone calls and sent email correspondence to Derek Yun at his known personal email address and personal telephone number on file with Lumos, providing identical notice.
(d) Marcus Delgado sent email correspondence and placed a telephone call to NovaSpark Technologies Inc. at its principal address at 88 Bluxome Street, Floor 4, San Francisco, California 94107, directed to the attention of Rajiv Anand in his capacity as a principal of NovaSpark.
As of the filing of this application, no counsel of record has appeared on behalf of any of the Defendants in this action. The Verified Complaint and this ex parte application, together with all supporting papers, have been served on each of the Defendants. Defendants may appear through counsel of their choosing at the hearing on this application or on the Order to Show Cause.
NovaSpark and the individual Defendants have been served with the Verified Complaint and this application by personal service and, where personal service was not immediately achievable, by substituted service in compliance with the California Code of Civil Procedure.
D. Accompanying Papers
This ex parte application is supported by and submitted concurrently with the following papers:
1. Verified Complaint for Trade Secret Misappropriation, Breach of Contract, Breach of Duty of Loyalty, Unfair Business Practices, and Conversion (filed concurrently, Case No. 23-CV-004871) (Exhibit A);
2. Declaration of Priya Chandrasekaran, Chief Executive Officer of Lumos Analytics Inc., with exhibits (Exhibit B);
3. Declaration of Rachel Fong, Vice President of Engineering of Lumos Analytics Inc., with exhibits (Exhibit C);
4. Memorandum of Points and Authorities in Support of Ex Parte Application for Temporary Restraining Order and Order to Show Cause Re: Preliminary Injunction (incorporated herein as Section IV);
5. [Proposed] Temporary Restraining Order (Exhibit D);
6. [Proposed] Order to Show Cause Re: Preliminary Injunction (Exhibit E).
II. INTRODUCTION
This case arises from a calculated and coordinated scheme by three former senior engineers of Lumos Analytics Inc. to steal the company's most valuable and closely guarded trade secrets and use them to launch a direct competitor — NovaSpark Technologies Inc. — virtually overnight.
Lumos Analytics Inc. is a Delaware-incorporated, California-headquartered software-as-a-service ("SaaS") company specializing in artificial-intelligence-driven financial analytics for institutional asset managers. Lumos's principal place of business is located at 4700 Orchard Parkway, Suite 300, San Jose, California 95134. Founded in approximately 2017, Lumos has raised a total of $185 million in venture capital funding through its Series D round, employs approximately 340 people, and generated approximately $67 million in annual recurring revenue ("ARR") in 2022. Lumos's flagship product, LumosEdge, is a proprietary platform that uses ensemble machine-learning models to generate predictive signals for equity and fixed-income portfolio optimization. LumosEdge serves some of the most sophisticated institutional investors in the world and represents the culmination of years of intensive research and development, millions of dollars of capital investment, and the combined expertise of Lumos's engineering and data-science teams.
Three former senior engineers — Defendants Rajiv Anand, Serena Mallick, and Derek Yun — occupied positions of trust and responsibility at Lumos. Anand served as Principal Engineer — ML Infrastructure and was a key architect of the LumosEdge signal-generation engine. Mallick served as Senior Software Engineer — Data Pipelines and was responsible for building and maintaining the proprietary DataStream ETL pipeline system. Yun served as Machine Learning Engineer II with direct access to the company's proprietary training datasets, model hyperparameters, and feature-engineering libraries. Together, these three individuals possessed access to virtually every component of Lumos's core proprietary technology stack.
Each of these individuals executed a Proprietary Information and Inventions Assignment Agreement ("PIIAA") upon commencing employment with Lumos. Each PIIAA contained comprehensive confidentiality and non-disclosure obligations, invention assignment provisions, non-solicitation covenants with a twelve-month post-separation duration, and express covenants not to use or disclose Lumos trade secrets at any time. The confidentiality obligations of the PIIAAs survive termination of employment for a period of three years.
In September and October 2022, Anand, Mallick, and Yun departed Lumos in rapid succession. The timing and sequence of their departures was anything but coincidental. In the days immediately preceding each resignation, each defendant engaged in systematic exfiltration of Lumos proprietary information. Anand transferred approximately 3.2 gigabytes of files — including model architecture specifications and hyperparameter configuration files for the LumosEdge signal-generation engine — to a personal Hollowell Health IT account, and forwarded the confidential 37-page LumosEdge 3.0 Product Roadmap to his personal email. Mallick cloned the entire DataStream ETL pipeline repository, approximately 780 megabytes in size, to a personal external USB drive. Yun downloaded 47 proprietary Jupyter notebooks containing feature-engineering code and model-training scripts to a personal Larkspur Technologies account. None of these transfers was authorized by Lumos, and each was conducted furtively and outside normal business workflows.
Just three days after Anand's resignation, on October 3, 2022, he incorporated NovaSpark Technologies Inc. in Delaware. NovaSpark purports to offer AI-driven alternative-data analytics for hedge funds and proprietary trading desks — a market directly and substantially overlapping with Lumos's core offering. Within days of their own departures, Mallick and Yun joined NovaSpark — Mallick as its Vice President of Engineering and Yun as a founding engineer. In November 2022, NovaSpark secured $12 million in seed funding from Blackpine Capital Partners LP, enabling Defendants to accelerate development of a competing product built upon Lumos's misappropriated technology.
Lumos discovered the exfiltration on December 12, 2022, when its IT security team identified anomalous data-transfer activity in the course of routine security monitoring. On January 9, 2023, Lumos sent cease-and-desist letters to each individual defendant and to NovaSpark, demanding the immediate return of all Lumos proprietary materials and cessation of any use or disclosure of Lumos trade secrets. As of the date of this filing, no substantive response has been received from any defendant.
The harm to Lumos is ongoing and accelerating with each passing day. NovaSpark is actively building a competing product, and the continued use and dissemination of Lumos's trade secrets deepens the injury with every line of code written, every model trained, and every investor presentation delivered. Absent emergency relief from this Court, Lumos's competitive advantage — painstakingly developed over years and at a cost of hundreds of millions of dollars — will be irretrievably lost. Monetary damages cannot restore the secrecy of trade secrets once disclosed, nor can they recapture the competitive position that Lumos stands to lose.
Lumos therefore seeks an ex parte Temporary Restraining Order to preserve the status quo and to prevent Defendants from continuing to exploit Lumos's proprietary information, destroying critical evidence, or further disseminating Lumos's trade secrets to additional third parties. Lumos further requests that the Court issue an Order to Show Cause Re: Preliminary Injunction, to be heard at the earliest available date, so that the Court may consider whether to extend the injunctive relief pending a full adjudication on the merits.
III. STATEMENT OF FACTS
A. Lumos Analytics Inc. and Its Proprietary Technology
1. Company Background
Lumos Analytics Inc. is a corporation organized under the laws of the State of Delaware (Delaware File No. 6518274; Employer Identification Number 82-4713956) with its principal place of business at 4700 Orchard Parkway, Suite 300, San Jose, California 95134. (Chandrasekaran Decl. ¶ 2.) Lumos was founded in approximately 2017 with the mission of applying advanced machine-learning techniques to the financial analytics domain, with a specific focus on generating predictive signals for institutional asset managers. (Id. ¶ 3.)
Lumos's Chief Executive Officer is Priya Chandrasekaran. Its General Counsel is Thomas Brightwell. Its Vice President of Engineering is Rachel Fong. (Chandrasekaran Decl. ¶ 4.) The company employs approximately 340 individuals, the majority of whom are based in its San Jose headquarters, with additional engineering staff distributed across offices in San Francisco and New York. (Id. ¶ 5.)
Since its founding, Lumos has attracted substantial venture capital investment reflecting the market's recognition of the value and promise of its proprietary technology. Lumos has raised a total of approximately $185 million through its Series D round of financing. Notable rounds include a $40 million Series B led by Ridgeline Ventures in 2019 and a $72 million Series D led by Northstar Growth Equity in June 2022. (Chandrasekaran Decl. ¶ 6.) In 2022, Lumos's annual recurring revenue was approximately $67 million, reflecting strong and growing demand for its LumosEdge platform among institutional investors, including hedge funds, pension funds, endowments, and family offices. (Id. ¶ 7.)
Lumos's flagship product is LumosEdge, a cloud-based analytics platform that uses proprietary ensemble machine-learning models to generate predictive signals for equity and fixed-income portfolio optimization. LumosEdge ingests data from dozens of alternative and traditional data sources, processes that data through a proprietary pipeline, engineers thousands of predictive features, and applies sophisticated ML models to generate signals that enable asset managers to make more informed investment decisions. (Fong Decl. ¶ 3.) LumosEdge represents the culmination of years of intensive research and development by a team of talented engineers and data scientists, and it is the core of Lumos's business and competitive position in the market. (Chandrasekaran Decl. ¶ 8.)
2. The Trade Secrets at Issue
At the heart of this action are five categories of proprietary information that constitute trade secrets of Lumos Analytics Inc. under the California Uniform Trade Secrets Act, California Civil Code section 3426.1(d). Each category derives independent economic value, both actual and potential, from not being generally known to the public or to other persons who can obtain economic value from its disclosure or use. Each category is the subject of efforts that are reasonable under the circumstances to maintain its secrecy. (Fong Decl. ¶¶ 5–20; Chandrasekaran Decl. ¶¶ 10–15.)
The five categories are as follows:
(i) LumosEdge Ensemble ML Model Architecture. The LumosEdge signal-generation engine is powered by a proprietary ensemble of machine-learning models. The specific neural-network topology, hyperparameter configurations, model architecture specifications, weighting schemes, and integration logic that comprise this ensemble are the product of years of iterative development, experimentation, and optimization by Lumos's ML engineering team. These specifications define how LumosEdge processes data and generates predictive signals, and they constitute the technical core of the platform. No component of this architecture has been published or disclosed outside of Lumos. (Fong Decl. ¶¶ 6–8.)
(ii) Proprietary Feature-Engineering Libraries. Lumos has developed a library of 2,847 individually engineered features for financial signal generation. Each feature represents a specific mathematical transformation or derivation applied to raw or processed data inputs, designed to capture predictive relationships relevant to asset pricing and portfolio construction. The development of this feature library required years of R&D investment, experimentation, and domain expertise. The specific features, their formulations, and their integration into the LumosEdge pipeline are not publicly known. (Fong Decl. ¶¶ 9–11.)
(iii) DataStream ETL Pipeline Codebase. The DataStream system is Lumos's proprietary extract, transform, and load ("ETL") pipeline. It governs the ingestion, normalization, validation, and routing of data from 43 alternative-data vendors — including satellite imagery providers, web-scraping services, transaction aggregators, and social-media analytics platforms — into the LumosEdge processing environment. The codebase includes data-schema definitions, connector specifications, error-handling logic, and quality-assurance modules. The DataStream system is critical to LumosEdge's ability to ingest diverse data sources at scale and with fidelity. (Fong Decl. ¶¶ 12–14.)
(iv) Curated Training Dataset. Lumos has assembled and curated a training dataset of approximately 14.6 terabytes comprising historical financial and alternative data, labeled and organized for the purpose of training the LumosEdge ML models. The curation process — which includes the selection, cleaning, labeling, augmentation, and organization of data — is itself a proprietary methodology that reflects years of effort and significant financial investment. The dataset is stored on Lumos's secured cloud infrastructure and is accessible only to authorized personnel. (Fong Decl. ¶¶ 15–17.)
(v) Confidential Product Roadmap. Lumos has developed a 37-page internal document titled "LumosEdge 3.0 Product Roadmap" that sets forth the company's planned product development for the period 2023–2025. This document describes planned new features, technical improvements, target markets, competitive positioning strategies, and projected timelines for release. The roadmap was developed by senior leadership and engineering personnel and was classified as strictly confidential. It has never been published or disclosed to any person outside of Lumos other than certain investors subject to confidentiality obligations. (Chandrasekaran Decl. ¶¶ 12–14; Fong Decl. ¶¶ 18–19.)
These five categories of proprietary information represent years of R&D investment, constitute the core competitive advantage of Lumos in the institutional financial analytics market, and are not publicly available or generally known to persons in the relevant industry. (Chandrasekaran Decl. ¶ 15.)
Lumos maintains robust security measures to protect these trade secrets. Access to the LumosEdge codebase, model architecture specifications, feature libraries, DataStream pipelines, and training datasets is restricted to authorized employees through role-based access controls, multi-factor authentication, and encrypted connections. All employees with access to proprietary information are required to execute the PIIAA as a condition of employment. Lumos's IT security infrastructure includes endpoint detection and response tools, data-loss-prevention software, network monitoring, and regular security audits. Physical access to Lumos's offices is controlled through badge-access systems. Sensitive documents are marked as confidential and subject to internal distribution controls. (Fong Decl. ¶ 20; Chandrasekaran Decl. ¶ 16.)
B. The Individual Defendants and Their Access to Trade Secrets
1. Rajiv Anand
Defendant Rajiv Anand, age 37, was employed by Lumos as Principal Engineer — ML Infrastructure from approximately March 2019 through September 30, 2022. (Chandrasekaran Decl. ¶ 18.) As Principal Engineer, Anand was one of the key architects of the LumosEdge signal-generation engine. He was involved in the design, development, testing, and optimization of the ensemble ML model architecture and had deep familiarity with the neural-network topology, hyperparameter configurations, and integration logic underlying the platform. (Id. ¶ 19; Fong Decl. ¶ 22.)
By virtue of his role, Anand had access to Lumos's most sensitive proprietary codebases and training data. Specifically, Anand had access to all five categories of trade secrets identified above: the LumosEdge ensemble ML model architecture, the proprietary feature-engineering libraries, the DataStream ETL pipeline codebase, the curated training dataset, and the confidential LumosEdge 3.0 Product Roadmap. (Fong Decl. ¶ 23.)
Upon commencing his employment with Lumos, Anand executed a Proprietary Information and Inventions Assignment Agreement dated March 11, 2019. The PIIAA contains, among other provisions: (a) an invention assignment clause assigning to Lumos all right, title, and interest in any inventions, developments, or improvements made during the course of employment; (b) a confidentiality and non-disclosure covenant prohibiting the use or disclosure of Lumos proprietary information during and after employment, with the confidentiality obligation surviving termination for three years; (c) a non-solicitation covenant prohibiting solicitation of Lumos employees for a period of twelve months following separation; and (d) a covenant not to use or disclose Lumos trade secrets at any time. A true and correct copy of Anand's executed PIIAA is attached as Exhibit 1 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 20, Ex. 1.)
Anand's residential address is 1482 Mariposa Avenue, Palo Alto, California 94306. (Chandrasekaran Decl. ¶ 21.)
2. Serena Mallick
Defendant Serena Mallick, age 34, was employed by Lumos as Senior Software Engineer — Data Pipelines from approximately June 2020 through October 14, 2022. (Chandrasekaran Decl. ¶ 22.) In this role, Mallick was responsible for building, maintaining, and enhancing the proprietary DataStream ETL pipelines that form the backbone of LumosEdge's data-ingestion system. (Id. ¶ 23; Fong Decl. ¶ 25.)
Mallick had direct access to the DataStream codebase, data-schema definitions, connector specifications, and related technical specifications. She also had access to portions of the broader LumosEdge system as required for integration testing and pipeline validation. (Fong Decl. ¶ 26.)
Upon commencing her employment, Mallick executed a PIIAA dated June 15, 2020, containing the same material provisions as Anand's PIIAA, including invention assignment, confidentiality and non-disclosure obligations surviving three years post-termination, a twelve-month post-separation non-solicitation covenant, and a covenant not to use or disclose Lumos trade secrets at any time. A true and correct copy of Mallick's executed PIIAA is attached as Exhibit 2 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 24, Ex. 2.)
Mallick's residential address is 578 Guerrero Street, Apartment 3B, San Francisco, California 94110. (Chandrasekaran Decl. ¶ 25.)
3. Derek Yun
Defendant Derek Yun, age 31, was employed by Lumos as Machine Learning Engineer II from approximately January 2021 through October 21, 2022. (Chandrasekaran Decl. ¶ 26.) In this role, Yun was responsible for model training, evaluation, and optimization, and had direct access to proprietary training datasets, model hyperparameters, and feature-engineering libraries. (Id. ¶ 27; Fong Decl. ¶ 28.)
Yun's work involved extensive use of proprietary Jupyter notebooks containing feature-engineering code and model-training scripts that embody Lumos's proprietary methodologies for constructing the 2,847 engineered features comprising the company's feature library. (Fong Decl. ¶ 29.)
Upon commencing his employment, Yun executed a PIIAA dated January 11, 2021, containing the same material provisions as the PIIAAs executed by Anand and Mallick. A true and correct copy of Yun's executed PIIAA is attached as Exhibit 3 to the Declaration of Priya Chandrasekaran. (Chandrasekaran Decl. ¶ 28, Ex. 3.)
Yun's residential address is 2241 Cowper Street, Apartment 12, Palo Alto, California 94301. (Chandrasekaran Decl. ¶ 29.)
C. The Coordinated Departures and Formation of NovaSpark
The sequence of events surrounding the individual Defendants' departures from Lumos and the formation of NovaSpark Technologies Inc. reveals a carefully orchestrated scheme to misappropriate Lumos's proprietary technology and deploy it at a new, directly competing venture. The relevant timeline is as follows:
September 30, 2022: Rajiv Anand resigns from Lumos, effective immediately, without providing the customary two-week notice period and without returning company materials or completing the standard offboarding procedures. Anand's resignation letter was terse and provided no explanation for his departure. (Chandrasekaran Decl. ¶ 30.)
October 3, 2022: Just three days after his resignation from Lumos, Anand incorporates NovaSpark Technologies Inc. in the State of Delaware (Delaware File No. 7294183; Employer Identification Number 88-3291047). NovaSpark establishes its headquarters at 88 Bluxome Street, Floor 4, San Francisco, California 94107. (Chandrasekaran Decl. ¶ 31, Ex. 4.)
NovaSpark's publicly stated business description indicates that it purports to offer "AI-driven alternative-data analytics for hedge funds and proprietary trading desks." This description identifies a market that directly and substantially overlaps with Lumos's core offering. The speed with which NovaSpark was incorporated — within days of Anand's departure — strongly suggests that the planning and preparatory work for the new venture was conducted while Anand was still employed by Lumos. (Chandrasekaran Decl. ¶ 32.)
October 14, 2022: Serena Mallick resigns from Lumos, effective immediately, without providing the customary notice period or completing standard offboarding procedures. (Chandrasekaran Decl. ¶ 33.)
October 17, 2022: Mallick begins employment at NovaSpark as Vice President of Engineering — only three days after her resignation from Lumos. Her Westmark Health IT profile was updated to reflect the NovaSpark role on or about this date. (Chandrasekaran Decl. ¶ 34.)
October 21, 2022: Derek Yun resigns from Lumos, effective immediately, without providing the customary notice period or completing standard offboarding procedures. (Chandrasekaran Decl. ¶ 35.)
October 24, 2022: Yun begins employment at NovaSpark as a founding engineer — only three days after his resignation from Lumos. (Chandrasekaran Decl. ¶ 36.)
November 15, 2022: NovaSpark closes a $12 million seed financing round from Blackpine Capital Partners LP (Managing Partner: Franklin Osei), a venture capital firm headquartered at 101 Mission Street, 38th Floor, San Francisco, California 94105. The closing of a $12 million seed round within approximately six weeks of NovaSpark's incorporation is unusual and suggests that NovaSpark was able to demonstrate a level of product development and technological sophistication that would ordinarily require many months, if not years, of work — a circumstance that is consistent with and strongly suggests the use of Lumos's misappropriated technology. (Chandrasekaran Decl. ¶ 37.)
The rapid sequence of departures, the immediate formation of a direct competitor, and the near-instantaneous hiring of all three departing engineers at the new entity are not coincidences. Taken together, these facts strongly support the inference that the individual Defendants engaged in a coordinated, premeditated scheme to misappropriate Lumos's trade secrets and use them to build NovaSpark. (Chandrasekaran Decl. ¶ 38.)
D. Evidence of Data Exfiltration
On December 12, 2022, Lumos's IT security team, in the course of routine security monitoring and a targeted post-departure audit of the individual Defendants' system activity, identified anomalous data-exfiltration activity dating to the final weeks of Anand's, Mallick's, and Yun's employment at Lumos. (Fong Decl. ¶ 31.) The internal investigation, conducted under the direction of VP of Engineering Rachel Fong and Lumos's Director of Information Security, Jonathan Mehta, revealed the following specific acts of misappropriation by each individual Defendant:
1. Rajiv Anand
On September 26, 2022 — four days before his resignation — Anand transferred approximately 3.2 gigabytes of files from Lumos's secure internal repositories to a personal Hollowell Health IT account. The transferred files included model architecture specifications, hyperparameter configuration files, and related documentation for the LumosEdge signal-generation engine. The transfer was identified through Lumos's data-loss-prevention ("DLP") system and confirmed through analysis of cloud-sync logs, endpoint monitoring data, and network egress records. (Fong Decl. ¶ 33, Ex. A.)
On September 28, 2022 — two days before his resignation — Anand accessed the confidential 37-page document titled "LumosEdge 3.0 Product Roadmap" from Lumos's internal document-management system and forwarded it as a PDF attachment to his personal email address (r.anand.personal@vaultmail.com). The email was captured by Lumos's email monitoring system, and the access and forwarding event is reflected in the system audit logs. (Fong Decl. ¶ 34, Ex. B.)
These transfers occurred outside Anand's normal business workflow. There was no legitimate business reason for Anand to transfer model architecture files or the product roadmap to personal, non-Lumos accounts. The transfers were not authorized by Lumos management, and no authorization was sought. (Fong Decl. ¶ 35.)
2. Serena Mallick
On October 10, 2022 — four days before her resignation — Mallick connected a personal external USB drive, identified by serial number WD-4837291, to her Lumos-issued workstation. She then cloned the entire DataStream ETL pipeline repository, comprising approximately 780 megabytes of proprietary source code, data-schema definitions, connector specifications, configuration files, and related documentation. The connection of the external device and the file-copy operation were recorded by Lumos's endpoint detection and response ("EDR") system. (Fong Decl. ¶ 37, Ex. C.)
The DataStream repository contains the full codebase and data-schema definitions for Lumos's proprietary data-ingestion system covering 43 alternative-data vendors. The repository is maintained in Lumos's private Git instance and is accessible only to authorized engineering personnel through authenticated, encrypted connections. (Fong Decl. ¶ 38.)
The cloning of the entire repository to a personal external device was not authorized by Lumos, was outside the scope of Mallick's normal duties, and was contrary to Lumos's information-security policies, which expressly prohibit the copying of proprietary source code to personal or unauthorized devices. (Fong Decl. ¶ 39.)
3. Derek Yun
On October 18, 2022 — three days before his resignation — Yun downloaded 47 proprietary Jupyter notebooks from Lumos's internal development environment to a personal Larkspur Technologies account. The downloaded notebooks contained feature-engineering code and model-training scripts that embody Lumos's proprietary methodologies for constructing the 2,847 engineered features that form the core of its feature-engineering library. The download was identified through cloud-sync agent logs on Yun's Lumos-issued laptop and confirmed through analysis of Larkspur Technologies API activity captured by the DLP system. (Fong Decl. ¶ 41, Ex. D.)
These Jupyter notebooks represent the practical implementation of Lumos's most closely guarded feature-engineering techniques. Their contents would enable a competitor to replicate or closely approximate the feature-engineering capabilities of LumosEdge without independently investing the years of R&D effort required to develop them. (Fong Decl. ¶ 42.)
The download was not authorized by Lumos. There was no legitimate business reason for Yun to transfer these notebooks to a personal Larkspur Technologies account, and no authorization was sought. (Fong Decl. ¶ 43.)
E. Failure to Return Company Materials
Under the terms of their respective PIIAAs, each individual Defendant was obligated to return all company materials, documents, data, equipment, and proprietary information upon the termination of their employment with Lumos. Section 5 of each PIIAA states, in pertinent part:
"Upon termination of my employment with the Company, I will promptly deliver to the Company, and will not keep in my possession, recreate, or deliver to anyone else, any and all devices, records, data, notes, reports, proposals, lists, correspondence, specifications, drawings, blueprints, sketches, laboratory notebooks, materials, flow charts, equipment, other documents or property, or reproductions of any of the aforementioned items developed by me pursuant to my employment with the Company or otherwise belonging to the Company, its successors, or assigns."
(Chandrasekaran Decl. ¶ 39, Exs. 1–3.)
Despite this clear contractual obligation, none of the three individual Defendants returned company materials upon their departure from Lumos. None of the three confirmed the deletion of Lumos proprietary information from their personal devices or accounts. Anand returned his Lumos-issued laptop in a perfunctory manner but did not return or account for any of the materials he had transferred to his personal Hollowell Health IT or personal email. Mallick returned her Lumos-issued laptop but did not return the external USB drive (serial number WD-4837291) containing the cloned DataStream repository or account for the data on it. Yun returned his Lumos-issued laptop but did not confirm deletion of the Jupyter notebooks from his personal Larkspur Technologies account. (Chandrasekaran Decl. ¶¶ 40–42; Fong Decl. ¶ 44.)
On January 9, 2023, Lumos, through its General Counsel Thomas Brightwell, sent cease-and-desist letters to each of the individual Defendants and to NovaSpark Technologies Inc. by certified mail and email. The letters demanded: (i) the immediate return of all Lumos proprietary materials in Defendants' possession, custody, or control; (ii) the immediate cessation of all use and disclosure of Lumos trade secrets and proprietary information; (iii) confirmation that all copies of Lumos proprietary information had been permanently deleted from all personal devices, accounts, and storage media; and (iv) a written response within seven days. (Chandrasekaran Decl. ¶ 43, Ex. 5.)
As of the date of this filing — January 19, 2023 — no substantive response has been received from any Defendant to the cease-and-desist letters. The silence of the Defendants in the face of these demands heightens the urgency of this application and the risk that Lumos's proprietary information is being used, disseminated, or destroyed. (Chandrasekaran Decl. ¶ 44.)
F. Retention of Counsel and Decision to Seek Emergency Relief
On December 15, 2022, following the discovery of the data-exfiltration activity, Lumos retained Calderwood & Harkness LLP to investigate the circumstances of the individual Defendants' departures and the formation of NovaSpark, and to evaluate and pursue all legal remedies available to Lumos. (Chandrasekaran Decl. ¶ 45.)
Based on its investigation — including review of the IT security findings, analysis of the PIIAAs, examination of publicly available information regarding NovaSpark, and consultations with Lumos's CEO and VP of Engineering — Calderwood & Harkness determined that emergency judicial relief is necessary and warranted. The ongoing nature of the misappropriation, the accelerating development of NovaSpark's competing product, the Defendants' failure to respond to the cease-and-desist letters, and the risk of further dissemination and destruction of evidence all counsel in favor of seeking a TRO on an ex parte basis. Every day of delay increases the irreparable harm to Lumos and diminishes the effectiveness of any eventual judicial remedy. (Chandrasekaran Decl. ¶ 46.)
IV. MEMORANDUM OF POINTS AND AUTHORITIES
A. Legal Standard for Ex Parte Temporary Restraining Order
California Code of Civil Procedure section 527(a) authorizes a court to grant a temporary restraining order upon a showing that "great or irreparable injury" will result to the applicant before the matter can be heard on notice. Cal. Code Civ. Proc. § 527(a). The purpose of a temporary restraining order is to preserve the status quo pending a hearing on a preliminary injunction. Oiye v. Fox (2012) 211 Cal.App.4th 1036, 1047.
California Rules of Court, Rule 3.1201 et seq., governs the procedural requirements for ex parte applications, including requirements for notice to the opposing party and the content of the application. The applicant must make an affirmative factual showing of the necessity for the temporary restraining order by a declaration containing specific facts showing that immediate and irreparable injury, loss, or damage will result if the relief is not granted before the opposing party or its attorney can be heard in opposition. Cal. Rules of Court, Rule 3.1202(c).
To obtain a temporary restraining order, the applicant must demonstrate: (1) a likelihood of prevailing on the merits at trial; (2) that the applicant will suffer irreparable harm in the absence of injunctive relief; (3) that the balance of hardships tips in the applicant's favor; and (4) that the threat of immediate and irreparable injury justifies ex parte relief without full adversarial hearing. White v. Davis (2003) 30 Cal.4th 528, 554; ReadyLink Healthcare v. Cotton (2005) 126 Cal.App.4th 1006, 1016.
In cases involving trade secret misappropriation, the California Uniform Trade Secrets Act ("CUTSA"), codified at California Civil Code sections 3426–3426.11, provides specific statutory authorization for injunctive relief. Section 3426.2(a) provides that "[a]ctual or threatened misappropriation may be enjoined." Cal. Civ. Code § 3426.2(a). This provision reflects the legislative determination that trade secret misappropriation is a category of harm particularly amenable to, and often requiring, injunctive relief. Courts have recognized that trade secret cases present "the classic case for the grant of an injunction" because "[t]he very nature of the secret . . . constitutes an interest that cannot be properly protected by an award of monetary damages." Readylink Healthcare, 126 Cal.App.4th at 1020 (citing DVD Copy Control Ass'n, Inc. v. Bunner (2003) 31 Cal.4th 864, 878).
As demonstrated below, Lumos satisfies each of these requirements.
B. Lumos Is Likely to Prevail on the Merits
1. First Cause of Action: Trade Secret Misappropriation (Cal. Civ. Code §§ 3426–3426.11)
Lumos is likely to prevail on its claim for trade secret misappropriation under the California Uniform Trade Secrets Act. The elements of a claim for trade secret misappropriation under CUTSA are: (a) the existence of a trade secret; and (b) misappropriation of that trade secret by the defendant. Cal. Civ. Code § 3426.1; Sargent Fletcher, Inc. v. Able Corp. (2003) 110 Cal.App.4th 1658, 1665.
Existence of Trade Secrets. A "trade secret" is defined as information that "[d]erives independent economic value, actual or potential, from not being generally known to the public or to other persons who can obtain economic value from its disclosure or use" and "[i]s the subject of efforts that are reasonable under the circumstances to maintain its secrecy." Cal. Civ. Code § 3426.1(d).
The five categories of Lumos proprietary information identified in Section III.A.2 above satisfy both prongs of this definition. First, each category derives independent economic value from not being generally known: the ML model architecture defines the core analytic capability of LumosEdge; the feature-engineering libraries represent thousands of individually developed predictive features; the DataStream pipeline codebase enables the ingestion of data from 43 vendors; the training dataset reflects years of curation; and the product roadmap reveals Lumos's strategic direction for the next three years. (Fong Decl. ¶¶ 5–20; Chandrasekaran Decl. ¶¶ 10–15.) These materials are not publicly available, have not been disclosed outside of Lumos (except under confidentiality obligations to investors), and would confer significant competitive advantage upon any competitor that obtained them. (Chandrasekaran Decl. ¶ 15.)
Second, Lumos has taken reasonable steps to maintain the secrecy of this information, including role-based access controls, multi-factor authentication, encrypted connections, endpoint detection and response systems, data-loss prevention software, physical access controls, mandatory PIIAAs for all employees with access to proprietary information, and internal policies restricting the distribution of confidential documents. (Fong Decl. ¶ 20; Chandrasekaran Decl. ¶ 16.) These measures are more than adequate to satisfy the "reasonable efforts" requirement under California law. See FLIR Systems, Inc. v. Parrish (2009) 174 Cal.App.4th 1270, 1279 (reasonable measures include confidentiality agreements, restricted access, and security protocols).
Misappropriation. "Misappropriation" includes, among other things, the acquisition of a trade secret by a person who knows or has reason to know that the trade secret was acquired by improper means, or the disclosure or use of a trade secret without express or implied consent by a person who used improper means to acquire knowledge of the trade secret, or who knew or had reason to know that knowledge of the trade secret was acquired by improper means. Cal. Civ. Code § 3426.1(b).
Here, the evidence establishes that each individual Defendant acquired Lumos trade secrets through improper means — namely, the unauthorized copying and transfer of proprietary information to personal accounts and devices in the days immediately preceding their resignations. (Fong Decl. ¶¶ 33–43.) These transfers were conducted furtively, outside the scope of the Defendants' normal duties, and in violation of Lumos's information-security policies. The subsequent use of these trade secrets in connection with NovaSpark — a direct competitor incorporated within days of Anand's departure — constitutes ongoing misappropriation through disclosure and use.
NovaSpark is itself liable for misappropriation to the extent it acquired Lumos trade secrets through the individual Defendants with knowledge, or reason to know, that the information was obtained by improper means. Given that NovaSpark was founded by Anand and staffed by Mallick and Yun — the very individuals who exfiltrated the trade secrets — NovaSpark's knowledge is imputed as a matter of law. See Ajaxo Inc. v. ETrade Grp., Inc.* (2005) 135 Cal.App.4th 21, 66.
The documentary evidence — including IT security logs, file-transfer records, email forwarding records, endpoint monitoring data, and cloud-sync logs — provides strong evidentiary support for Lumos's misappropriation claim. (Fong Decl. ¶¶ 31–43, Exs. A–D.)
2. Second Cause of Action: Breach of Contract (PIIAA)
Lumos is also likely to prevail on its claim for breach of the PIIAAs executed by each individual Defendant. Each PIIAA constitutes a valid and enforceable contract between the individual Defendant and Lumos. Each PIIAA contains express obligations including: (a) a duty to maintain the confidentiality of Lumos proprietary information and not to use or disclose it during or after employment; (b) a duty to assign to Lumos all inventions made during the course of employment; (c) a duty to return all company materials upon termination; and (d) a non-solicitation covenant of twelve months' duration. (Chandrasekaran Decl. ¶¶ 20, 24, 28, Exs. 1–3.)
Each individual Defendant breached the PIIAA in multiple respects. First, each Defendant copied and removed Lumos proprietary information — model architecture files, pipeline code, Jupyter notebooks, and the product roadmap — in violation of the confidentiality and non-disclosure provisions. Second, each Defendant failed to return all company materials upon termination, as required by Section 5 of the PIIAA. Third, each Defendant has used and/or disclosed Lumos trade secrets in connection with NovaSpark, in violation of the covenant not to use or disclose Lumos trade secrets at any time.
The existence of the contracts, the specific terms breached, and the specific acts of breach are all supported by documentary evidence. (Chandrasekaran Decl. ¶¶ 39–44; Fong Decl. ¶¶ 31–44.)
3. Third Cause of Action: Breach of Duty of Loyalty
Under California law, an employee owes a duty of loyalty to his or her employer during the period of employment. This duty prohibits an employee from acting in competition with the employer, diverting business opportunities, or misappropriating the employer's property for personal benefit while the employment relationship subsists. Huong Que, Inc. v. Luu (2007) 150 Cal.App.4th 400, 410; Stokes v. Dole Nut Co. (1995) 41 Cal.App.4th 285, 295.
While still employed by Lumos, each individual Defendant engaged in acts of data exfiltration in preparation for competing with Lumos through NovaSpark. Anand's transfers on September 26 and 28, 2022, occurred during his employment at Lumos. Mallick's cloning of the DataStream repository on October 10, 2022, occurred during her employment at Lumos. Yun's download of Jupyter notebooks on October 18, 2022, occurred during his employment at Lumos. (Fong Decl. ¶¶ 33–43.)
These acts — taking proprietary information for the purpose of launching or supporting a competing venture — constitute clear breaches of the common-law duty of loyalty owed by each Defendant to Lumos during the period of employment. See Fowler v. Varian Assocs., Inc. (1987) 196 Cal.App.3d 34, 41 (employees may not misappropriate employer property for use in a competing business while still employed).
4. Fourth Cause of Action: Unfair Business Practices (Cal. Bus. & Prof. Code § 17200)
California Business and Professions Code section 17200 prohibits "any unlawful, unfair, or fraudulent business act or practice." The statute is broadly construed to encompass a wide range of unfair competitive conduct. Cel-Tech Commc'ns, Inc. v. Los Angeles Cellular Tel. Co. (1999) 20 Cal.4th 163, 180.
The Defendants' misappropriation of Lumos's trade secrets, breach of their contractual obligations under the PIIAAs, and breach of their duty of loyalty constitute unlawful business practices within the meaning of section 17200, because these acts violate the California Uniform Trade Secrets Act, applicable contract law, and the common-law duty of loyalty. See Glenrock Corporation (2003) 29 Cal.4th 1134, 1143 (section 17200 "borrows" violations of other laws and treats them as unlawful practices).
Moreover, NovaSpark's use of Lumos's proprietary technology to develop a competing product and gain an unfair competitive advantage constitutes an independently actionable unfair business practice. A business practice is "unfair" under section 17200 if it offends an established public policy or is immoral, unethical, oppressive, unscrupulous, or substantially injurious to consumers or competitors. Cel-Tech, 20 Cal.4th at 186–187. Building a competing business on stolen trade secrets satisfies this standard.
5. Fifth Cause of Action: Conversion
Conversion is "the wrongful exercise of dominion over the property of another." Lee v. Hanley (2015) 61 Cal.4th 1225, 1240. The elements of conversion are: (1) ownership or right to possession of the property; (2) wrongful disposition of the property right; and (3) damages. Id.
California courts have recognized that trade secrets and proprietary information constitute property interests that can be the subject of a conversion claim. See Fremont Indem. Co. v. Fremont Gen. Corp. (2007) 148 Cal.App.4th 97, 125 (conversion can apply to intangible property that is merged into a document or other tangible thing). The Defendants converted Lumos's proprietary information — including tangible embodiments thereof in the form of electronic files, code, and documents — by copying, taking, retaining, and using that information without authorization and contrary to Lumos's property rights. Each Defendant exercised dominion and control over Lumos's property in a manner inconsistent with Lumos's rights of ownership.
C. Lumos Will Suffer Irreparable Harm Absent Injunctive Relief
Trade secret misappropriation, by its very nature, causes irreparable harm. Once a trade secret is disclosed or used by an unauthorized person, the secrecy that gives it value is destroyed and cannot be restored. FLIR Systems, Inc. v. Parrish (2009) 174 Cal.App.4th 1270, 1278 ("[T]he disclosure of a trade secret in violation of a confidential relationship can constitute irreparable injury."); Readylink Healthcare, Inc. v. Cotton (2005) 126 Cal.App.4th 1006, 1024 (finding irreparable harm where trade secrets were being used by a competitor). California Civil Code section 3426.2 itself reflects the legislature's recognition that injunctive relief is the appropriate remedy for actual or threatened misappropriation, precisely because monetary relief is often inadequate. Cal. Civ. Code § 3426.2(a).
Here, the irreparable harm to Lumos is neither speculative nor remote — it is ongoing and accelerating. NovaSpark is actively developing a competing product using Lumos's misappropriated trade secrets. Every day that passes without judicial intervention, the Defendants continue to integrate Lumos's proprietary technology — its model architectures, feature libraries, pipeline code, and strategic product plans — into NovaSpark's systems. The deeper that integration proceeds, the more difficult it becomes to disentangle and the more completely Lumos's competitive advantage is destroyed.
The risk of further dissemination compounds the harm. NovaSpark is not a solo operation; it has received $12 million in funding from Blackpine Capital Partners LP and presumably has or will hire additional employees. Each additional person who gains access to Lumos's trade secrets widens the circle of dissemination and diminishes the secrecy — and therefore the value — of the information. Investor presentations, recruiting materials, and customer demonstrations may further expose Lumos's proprietary technology to parties beyond the Defendants' immediate control.
Monetary damages are inadequate to redress the harm for at least three reasons. First, the value of trade secrets is inherently difficult to quantify once the secrecy has been breached; the harm is not merely the cost of developing the information but the permanent loss of the competitive advantage it confers. (Chandrasekaran Decl. ¶ 47.) Second, Lumos's competitive position — built over years of effort and hundreds of millions of dollars in investment — cannot be recaptured through a damages award after the fact; a competitor that reaches the market with a product built on Lumos's technology will permanently alter the competitive landscape. (Id. ¶ 48.) Third, NovaSpark's accelerated time-to-market, enabled by the misappropriated technology, causes market damage that compounds over time as NovaSpark acquires customers, builds relationships, and establishes brand recognition at Lumos's expense. (Id. ¶ 49.)
Accordingly, Lumos will suffer immediate and irreparable harm absent the issuance of a temporary restraining order.
D. The Balance of Hardships Favors Lumos
The balance of hardships tips decisively in favor of Lumos.
Harm to Lumos absent relief: Lumos stands to lose its core competitive advantage — proprietary technology representing years of research and development and hundreds of millions of dollars of capital investment. LumosEdge is the foundation of Lumos's business and the basis on which it has attracted customers, investors, and employees. If Defendants are permitted to continue using Lumos's trade secrets to build NovaSpark, Lumos's competitive position in the institutional asset management analytics market will be permanently and irreparably impaired. (Chandrasekaran Decl. ¶¶ 47–49.)
Burden on Defendants if relief is granted: The proposed TRO imposes only the requirement that Defendants refrain from using property that does not belong to them and that they preserve evidence relevant to this litigation. The TRO does not prevent NovaSpark from operating as a business, does not prevent the individual Defendants from pursuing their careers, and does not prohibit NovaSpark from competing with Lumos using independently developed technology or publicly available information. The TRO merely requires Defendants to cease exploiting Lumos's stolen trade secrets — an obligation they already owe under their PIIAAs and under the law. Defendants can articulate no legitimate interest in continuing to use property that was acquired through wrongful conduct. See Bramwell Partners LLC v. Hamidi (2003) 30 Cal.4th 1342, 1352 (no cognizable interest in continuing an unlawful act).
The hardship to Defendants of complying with a narrowly tailored restraining order is minimal compared to the hardship to Lumos of losing its trade secrets permanently. The equities overwhelmingly favor Lumos.
E. The Circumstances Justify Ex Parte Relief
Ex parte relief is warranted when there is an immediate threat that the opposing party will destroy evidence or cause further irreparable harm if given advance notice of the application. Cal. Rules of Court, Rule 3.1202(c); see also O'Connell v. Superior Court (2006) 141 Cal.App.4th 1452, 1463.
Here, there is a concrete and immediate threat that Defendants will destroy evidence or further disseminate Lumos's trade secrets if they are given advance notice of the specific terms of the relief sought. The Defendants have already demonstrated a willingness to misappropriate confidential information through covert means — transferring files to personal accounts, cloning repositories to external drives, and forwarding documents to personal email addresses. Individuals who have shown this disregard for their obligations cannot be trusted to preserve the very evidence that proves their wrongdoing.
Moreover, the cease-and-desist letters sent by Lumos on January 9, 2023, demanded the return of materials and cessation of use. The Defendants' failure to respond — ten days of silence — heightens the risk of spoliation. A party that ignores a cease-and-desist demand is more likely, not less, to delete or conceal evidence when it learns that litigation is imminent and that specific forensic inquiries will be directed at its devices and accounts.
Time is of the essence. NovaSpark is actively building its product. The longer Defendants are permitted to use Lumos's trade secrets, the more deeply those secrets become embedded in NovaSpark's systems, the more broadly they are disseminated, and the more difficult it will be to fashion effective relief. Delay in obtaining a TRO will only deepen the harm to Lumos and undermine the Court's ability to preserve the status quo. The urgency of the circumstances fully justifies the ex parte procedure.
F. The Proposed TRO Is Narrowly Tailored
The proposed TRO is narrowly tailored to address the specific harm at issue. The injunctive provisions address only Lumos's proprietary information and trade secrets. They do not prevent Defendants from conducting lawful business activities, pursuing employment in any field, or developing and marketing technology that is independently created or derived from publicly available sources.
The preservation requirements are similarly limited in scope. They require Defendants to maintain the status quo with respect to devices and accounts that may contain Lumos proprietary information. Preservation is necessary both to maintain the evidentiary record for this litigation and to enable forensic examination of Defendants' devices and accounts to determine the full scope of the misappropriation. Without a preservation order, there is a significant risk that critical evidence will be lost — whether through intentional deletion or through the operation of automatic data-retention and destruction policies.
The proposed TRO does not impose any bond requirement beyond the statutory minimum, and Lumos is prepared to post a bond in such amount as the Court may determine is appropriate.
In sum, the proposed TRO is the minimum relief necessary to preserve the status quo, protect Lumos's trade secrets, and maintain the integrity of the evidentiary record pending a full hearing on the preliminary injunction.
V. DESCRIPTION OF PROPOSED TEMPORARY RESTRAINING ORDER
Lumos respectfully submits the proposed Temporary Restraining Order attached hereto as Exhibit D. The key provisions of the proposed order are as follows:
A. Injunctive Provisions
Upon issuance of the TRO, Defendants — and each of them, and their officers, directors, agents, employees, representatives, successors, assigns, and all persons acting in concert or participation with them who receive actual notice of the order — shall immediately cease all use, disclosure, dissemination, distribution, publication, or exploitation of any and all Lumos Analytics Inc. trade secrets and proprietary information, including but not limited to:
(i) The LumosEdge ensemble ML model architecture, including all neural-network topology specifications, hyperparameter configurations, model architecture documents, weighting schemes, and integration logic;
(ii) The proprietary feature-engineering libraries, including all 2,847 individually engineered features, their formulations, and their integration specifications;
(iii) The DataStream ETL pipeline codebase, including all source code, data-schema definitions, connector specifications, configuration files, error-handling logic, and quality-assurance modules;
(iv) The curated training datasets, consisting of approximately 14.6 terabytes of historical financial and alternative data assembled and curated by Lumos for ML model training;
(v) The LumosEdge 3.0 Product Roadmap and all related confidential product development materials, including any excerpts, summaries, or derivative documents.
B. Preservation Provisions
Defendants shall immediately preserve and maintain in their current state all electronic devices, storage media, servers, cloud storage accounts, email accounts, and any and all repositories that contain or may contain Lumos proprietary information. Without limiting the generality of the foregoing, the specific devices and accounts to be preserved include but are not limited to:
(a) Rajiv Anand's personal Hollowell Health IT account (the account to which approximately 3.2 GB of files were transferred on September 26, 2022) and personal email account (r.anand.personal@vaultmail.com);
(b) Serena Mallick's personal external USB drive, serial number WD-4837291, and any copies, backups, or images of the data contained thereon;
(c) Derek Yun's personal Larkspur Technologies account (the account to which 47 Jupyter notebooks were downloaded on October 18, 2022);
(d) All NovaSpark Technologies Inc. servers, cloud infrastructure, code repositories (including but not limited to Git repositories), development environments, staging environments, production environments, and associated backup systems;
(e) All personal laptops, desktop computers, smartphones, tablets, external hard drives, USB storage devices, and any other el

SUPERIOR COURT OF THE STATE OF CALIFORNIA

COUNTY OF SANTA CLARA

UNLIMITED CIVIL JURISDICTION

TABLE OF CONTENTS

I. NOTICE OF EX PARTE APPLICATION

II. INTRODUCTION

III. STATEMENT OF FACTS

IV. MEMORANDUM OF POINTS AND AUTHORITIES

V. DESCRIPTION OF PROPOSED TEMPORARY RESTRAINING ORDER

Record

Parties

Dates

Amounts