Hudson Valley Biometrics, Inc. v. Priya Iyer

CASE-FILE DOCUMENT · EMPLOYMENT AGREEMENT

Employment Agreement (Excerpt) — Hudson Valley Biometrics, Inc.

Exhibit m16.exh.001 Employee: Priya Iyer Effective Date: June 3, 2019 Position: Algorithms Engineer

The following are the restrictive-covenant and confidentiality provisions of the standard-form Employment Agreement signed by the Employee as a condition of employment. Formatting and numbering are reproduced from the executed agreement.

7. Confidential Information and Trade Secrets. During and after employment, Employee shall hold in strict confidence and shall not use or disclose any Confidential Information of the Company. "Confidential Information" means all non-public information relating to the Company's technology, software, algorithms, models, source code, training data, methods, customers, pricing, and business plans, including without limitation the Company's proprietary person-matching systems and the methods by which they are designed, trained, and operated. Employee acknowledges that such information constitutes trade secrets of the Company.

8. Non-Competition. For a period of twenty-four (24) months following the termination of employment for any reason, Employee shall not, directly or indirectly, own, manage, be employed by, consult for, or provide services to any business engaged in biometric identification, computer-vision matching, or related fields. This restriction shall apply without geographic limitation, the parties acknowledging that the Company's business and its competitors operate without regard to geography.

9. Non-Solicitation. For a period of twenty-four (24) months following termination, Employee shall not, directly or indirectly, solicit, recruit, or induce any employee, contractor, customer, or prospective customer of the Company to terminate or curtail their relationship with the Company.

10. Acknowledgment. Employee acknowledges that the restrictions in Sections 7–9 are reasonable and necessary to protect the Company's legitimate business interests, that Employee's services are unique, and that any breach will cause the Company irreparable harm for which money damages would be inadequate, entitling the Company to injunctive relief.

11. At-Will Employment. Nothing in this Agreement alters the at-will nature of Employee's employment, which may be terminated by either party at any time, with or without cause.

12. Blue-Pencil. If any provision of Sections 7–9 is held unenforceable, a court may modify or limit such provision to the extent necessary to render it enforceable and shall enforce it as so modified.

AGREED:

/s/ Priya Iyer — June 3, 2019 For Hudson Valley Biometrics, Inc.: /s/ M. Calloway, VP People Operations — June 3, 2019

Handwritten note in the client's file copy, initialed "PI": "signed on first day w/ everyone else — never read closely."

CASE-FILE DOCUMENT · RESIGNATION NOTICE

Resignation Notice (Email) — Exhibit m16.exh.002

From: Priya Iyer To: Roland Steinhauer; M. Calloway (People Operations) Date: January 12, 2026, 9:14 AM Subject: Resignation — Priya Iyer

Roland and team,

I'm writing to give notice of my resignation from Hudson Valley Biometrics. My last day will be January 30, 2026, which gives us just over two weeks to hand off my work.

This was not an easy decision. I've spent nearly seven years here and I'm proud of what our team built together. But after the direction things took last year, I've concluded it's time for a change, and I've accepted a role elsewhere.

I want the transition to be clean. Over the next two weeks I'll document the open items on the matching-engine roadmap, write up the state of the current training runs, and make sure Dominic and the rest of the team have everything they need to keep moving. I'll return my laptop, phone, and badge on my last day, and I'm happy to sit for an exit interview.

I've valued working with all of you, and I'd like to leave on good terms.

Best, Priya

People Operations note appended to file copy: Devices (laptop, phone, badge) collected on January 30, 2026. IT security off-boarding review initiated same day.

CASE-FILE DOCUMENT · SIGHTLINE OFFER LETTER

Offer Letter and Job Description — Exhibit m16.exh.003

SIGHTLINE PERCEPTION, INC. Date: December 18, 2025

Ms. Priya Iyer

Dear Priya,

On behalf of Sightline Perception, Inc., I am delighted to offer you the position of Principal Engineer, Gait Access Platform, reporting to me. We were impressed by your depth in computer vision and by your enthusiasm for the access-control problem, which is a very different animal from large-scale identity search.

Role. You will lead the architecture of a new gait-recognition model for our access-control product line — software that lets buildings and secure facilities admit enrolled, authorized people by the way they walk. This is greenfield work: you will be designing our approach from the ground up.

Start date. February 9, 2026.

Compensation. Base salary of $228,000, an annual performance bonus target of 15%, and an equity grant to be documented in a separate agreement.

Conditions. This offer is contingent on the following, which we ask you to acknowledge in writing:

1. No prior-employer materials. You will bring with you no code, documents, data, or other materials belonging to or derived from any prior employer, including Hudson Valley Biometrics. Your work at Sightline will be built entirely from your own general skill and from Sightline and public resources.
2. Ethical wall. For an initial period, you will be walled off from any Sightline project involving face recognition or any work resembling your prior employer's products. Your access and assignments during this period will be documented.
3. Cooperation on restrictive covenants. You have disclosed to us that you signed a non-competition agreement with Hudson Valley Biometrics. We have reviewed it with counsel and are prepared to proceed, and to support you if your prior employer challenges your move.

We are excited to have you build this with us.

Sincerely,

/s/ Beatriz Salcedo Vice President of Engineering, Sightline Perception, Inc.

ACKNOWLEDGED AND ACCEPTED: /s/ Priya Iyer — December 20, 2025

Job Description — Principal Engineer, Gait Access Platform

• Architect and lead development of a novel gait-recognition model for physical access control.
• Target deployment: enrolled-user recognition at doorways and secure entries (small, cooperative, opt-in populations).
• Customers: building-security integrators, logistics and warehouse operators, facilities managers.
• Out of scope: face recognition; large-database identity search; law-enforcement or surveillance watch-list matching.

CASE-FILE DOCUMENT · MATCHING ENGINE DESCRIPTION

Description of the "Matching Engine" (Alleged Trade Secret) — Exhibit m16.exh.005

This exhibit is a plain-language description prepared for the file, drawn from the parties' submissions and from engineer interviews, of the system Hudson Valley Biometrics identifies as the trade secret at the center of its motion. No source code is reproduced. The description is offered so the reader can evaluate the protectability and inevitable-disclosure questions; it states no legal conclusions.

What Hudson Valley calls "the matching engine"

Hudson Valley Biometrics's flagship product converts video or images of a person's face or walking gait into a compact list of numbers — an "embedding" — and then compares that embedding against a stored database to find the closest matches. The company calls the software that does this its "matching engine." In its motion papers, Hudson Valley describes the engine as "the culmination of years of proprietary research and investment" and "the crown jewel of the Company," and it identifies Ms. Iyer's knowledge of the engine as the trade secret it seeks to protect.

The components, in plain terms

• The embedding model. A neural network that turns an image into a numeric vector. The network architecture is a variant of designs published in the academic computer-vision literature.
• The training method. The model is trained using loss functions and data-augmentation techniques that are described in published papers and implemented in widely used open-source machine-learning libraries.
• The matching step. Once an embedding is produced, the system finds the nearest entries in the database using standard nearest-neighbor search techniques.
• The proprietary layer. What Hudson Valley developed itself sits on top of these public foundations: the specific hyperparameters chosen through experimentation, the curated and labeled training data the company assembled, and the accumulated engineering judgment about what configurations work best for its customers' conditions.

How the engine was maintained internally (per engineer interviews)

• The source code resided in shared internal repositories accessible to the full engineering team.
• The architecture was presented and discussed openly in company engineering meetings.
• There was no separate password barrier or "need to know" compartment isolating the engine's core.
• No single document consistently marked "trade secret" described the engine as a whole; sensitivity was understood by custom rather than by formal marking.

What is disputed

Hudson Valley contends that Ms. Iyer, as the engine's technical lead, carries its trade secrets in her memory and will inevitably use or disclose them in any competing role. Ms. Iyer contends that the engine's methods are public, that what is genuinely proprietary (the tuning and data) she did not take in any form, that the company's off-boarding forensic review found no company materials on her devices, and that general skill and knowledge she developed over her career are hers to use.

CASE-FILE DOCUMENT · IT FORENSIC REVIEW

IT Off-Boarding Forensic Review and Device-Return Receipt — Exhibit m16.exh.006

HUDSON VALLEY BIOMETRICS, INC. — Information Security Off-Boarding Review Report

Departing Employee: Priya Iyer Last Day: January 30, 2026 Review Period: January 16 – January 31, 2026 Prepared by: T. Nagata, IT Security Analyst Report Date: February 4, 2026

1. Devices returned

All company-issued devices were returned on the employee's last day. No devices are outstanding.

2. Scope of review

Per standard practice for departing engineering staff, Information Security imaged the returned laptop and reviewed: (a) the employee's company email account for external-forwarding activity in the 60 days before departure; (b) company cloud-storage and code-repository access logs; (c) removable-media (USB) connection logs on the laptop; and (d) any personal cloud accounts the employee had linked to company systems.

3. Findings

• Source code / repositories. No evidence that company source code, model files, or training data were copied to any removable media, personal cloud account, or personal email. USB connection logs show no mass-storage device attached in the review period.
• Email forwarding. One outbound message from the employee's company account to her personal email address was identified, dated January 27, 2026, with an attached compressed folder. Security review of the attachment identified it as containing personal files (résumé drafts, conference presentation slides) and one hand-drawn architecture diagram. The diagram is not source code and not a controlled document; its status is noted for management's attention.
• Customer data. No customer records or contact files were found to have been exported.
• Bulk download. No anomalous bulk-download activity was detected in the review period.

4. Analyst note

With the exception of the single personal-folder email noted in Section 3, this review found no indication that company code, data, or customer information was exfiltrated. The forwarded folder has been preserved. This report states facts observed in the logs and images; it does not opine on any legal question.

/s/ T. Nagata, IT Security Analyst

CASE-FILE DOCUMENT · CEASE AND DESIST LETTER

Cease-and-Desist Letter — Exhibit m16.exh.004

HALBROOK RENNER & ISELIN LLP Attorneys at Law Albany, New York

Date: February 2, 2026 By Email and Certified Mail

Ms. Priya Iyer

Re: Breach of Restrictive Covenants — Hudson Valley Biometrics, Inc.

Dear Ms. Iyer:

We represent Hudson Valley Biometrics, Inc. ("HVB"). We write regarding your imminent employment with Sightline Perception, Inc., a direct competitor of HVB, and your obligations under the Employment Agreement you signed on June 3, 2019.

As you know, that Agreement contains a two-year non-competition covenant barring you from working for any business engaged in biometric identification or computer-vision matching; a two-year non-solicitation covenant; and a confidentiality covenant protecting HVB's trade secrets. As the former technical lead of HVB's proprietary matching engine, you possess intimate knowledge of the Company's most valuable trade secrets. HVB's position is that you cannot perform any role at a competitor such as Sightline without inevitably using or disclosing that confidential information, regardless of whether you copied any files.

Accordingly, HVB demands that you: (1) immediately cease and desist from commencing or continuing any employment with Sightline Perception or any other competitor for the duration of your non-competition covenant; (2) return to HVB any Company materials in your possession, in any form; and (3) confirm in writing, within five (5) business days of this letter, that you will comply.

Should you fail to provide the requested assurances, HVB is prepared to seek immediate injunctive relief in court, together with all other remedies available to it, including its costs and attorneys' fees. HVB takes the protection of its trade secrets and the enforcement of its agreements extremely seriously.

Govern yourself accordingly. This letter is sent without prejudice to any of HVB's rights, all of which are expressly reserved.

Very truly yours,

/s/ Gerald T. Iselin HALBROOK RENNER & ISELIN LLP Counsel for Hudson Valley Biometrics, Inc.

WITNESS STATEMENT · WITNESS FRAWLEY

Witness Statement — Dominic Frawley

Witness: Dominic Frawley, Senior Engineer, Hudson Valley Biometrics, Inc. Taken by: Ellingboe & Ravndal LLP Date: April 14, 2026

My name is Dominic Frawley. I'm a senior engineer at Hudson Valley Biometrics, and I still work there. I've been on the engineering team for about five years, and for most of that time I worked alongside Priya Iyer on the matching engine. I'm giving this statement voluntarily because I don't think what's being said about Priya — that she walked out with some secret algorithm in her head — is a fair description of how any of this actually works.

Let me start with what the "matching engine" is, because I think that's where the whole case goes sideways. It's not a magic box. It's a pipeline: you take an image or a video of someone, you run it through a neural network that turns it into a list of numbers — we call that an embedding — and then you compare those numbers against a database to find the closest matches. Every piece of that is standard computer-vision work. The network architectures we use come out of published academic papers. The training methods are in the literature. A lot of the actual code sits on top of open-source libraries that anyone can download for free. If you put ten qualified vision engineers in a room and described the problem, most of them would reach for the same tools.

What's genuinely ours is the tuning — the specific settings, the training data we've collected, the thousand small judgment calls that make our version work well. That part has real value. But that's not something you carry out in your head as a single secret. It's spread across a lot of people and a lot of files, and Priya didn't take any of those files. I know, because I still have access to all of them, and nothing's missing.

I also want to be honest about how we handled the code, because I keep hearing it described as locked down, and it wasn't. The source lived in shared repositories that the whole engineering team could read and check out. We talked about the architecture openly in engineering meetings — Priya used to present on it. There was no separate password wall around the "secret" parts, no compartment where only a few people could look. That's just how a fast-moving startup builds software. I'm not criticizing the company; I'm telling you it wasn't run like a classified program.

Priya was the best engineer I've worked with, and she was straight with people. When leadership passed her over last year, a lot of us thought it was a mistake, and I wasn't surprised she left. I don't know anything about what she emailed herself or what she took in a notebook — I can't speak to that. I can only tell you how the engine was built and how the code was treated while I was sitting three desks away from her, which is: like ordinary engineering work resting on public foundations, not like a vault.

I understand I may be asked to say this under oath, and I'm prepared to.

/s/ Dominic Frawley

WITNESS STATEMENT · WITNESS SALCEDO

Witness Statement — Beatriz Salcedo

Witness: Beatriz Salcedo, Vice President of Engineering, Sightline Perception, Inc. Taken by: Ellingboe & Ravndal LLP Date: April 21, 2026

My name is Beatriz Salcedo. I am the Vice President of Engineering at Sightline Perception, and I made the decision to hire Priya Iyer. I want to be clear at the outset about what we hired her to do, because I think there is a serious misunderstanding about it.

Sightline builds gait-based access control. In plain terms, our software recognizes authorized people by the way they walk, so that a building or a secure facility can let the right people through without badges or cards. Our customers are building-security companies, logistics operators, and facility managers. We are not in the law-enforcement market. We do not build face-recognition watch-list systems. We do not sell to police agencies. That is a different product for a different buyer, and it is squarely what Hudson Valley Biometrics does and we do not.

We hired Priya to architect a brand-new gait model from the ground up. She is not being asked to reproduce anything from her prior job, and the technical problem is genuinely different — recognizing a small set of enrolled, cooperative people walking through a doorway is not the same problem as searching a face against a large database. I understood when we made the offer that she had signed a non-compete, and we took that seriously. We are not interested in anyone else's trade secrets; frankly, they would not help us, because we are solving a different problem.

Before her start date we put specific safeguards in place. We instructed her in writing to bring no materials of any kind from Hudson Valley — no code, no documents, no notes derived from their systems — and she confirmed she had none. We set up what we call an ethical wall: for an initial period she is walled off from any project that touches face recognition or anything resembling her prior employer's work, and her onboarding was documented so that we could show exactly what she did and did not have access to. We did this precisely so that no one could later claim she brought something over.

When Hudson Valley sent its demand and then filed for an injunction, we agreed with Priya to pause her start date rather than press ahead while this is unresolved. I will say candidly that I found the timing striking. We had recently won a county public-safety bid that Hudson Valley had competed for, and the aggressiveness of their reaction to one engineer changing jobs seemed out of proportion to any real risk to their technology.

I am willing to confirm all of this under oath and to produce our onboarding and ethical-wall documentation if it is needed.

/s/ Beatriz Salcedo

WITNESS STATEMENT · WITNESS VANDERMEER

Witness Statement — Naomi Vandermeer

Witness: Naomi Vandermeer, former Machine-Learning Engineer, Hudson Valley Biometrics, Inc. Taken by: Ellingboe & Ravndal LLP Date: April 28, 2026

My name is Naomi Vandermeer. I worked at Hudson Valley Biometrics as a machine-learning engineer for about four years, and I left in 2024 to take a job at a company that has nothing to do with biometrics. I worked near Priya Iyer for most of my time there, and I'm giving this statement because I went through a version of what she's going through now, and I think that context matters.

First, about the non-compete. Everybody signed it. It was in the stack of onboarding paperwork on day one, along with the tax forms and the benefits elections. Nobody sat you down and explained it, and it was very clear you couldn't negotiate it — it was take-it-or-you-don't-have-the-job. I don't think a single engineer I knew there had a lawyer look at it before signing. It was the same two-year, no-boundaries language for everyone, whether you were a junior engineer or a lead. It read to me less like something tailored to protect a real secret and more like a standard scare-off clause.

Second, about what happened when I left. When I gave notice, Roland Steinhauer himself reminded me — pointedly — that I had signed a non-compete and that the company "took it very seriously." It felt like a warning. But I was going somewhere unrelated, and in the end they never did anything. No letter, no lawsuit, nothing. I've watched a few other people leave over the years, and as far as I know Hudson Valley has never actually sued anyone over that covenant until now. That's part of why the suit against Priya surprised me.

Third, about the "secret algorithm." I want to be careful because I'm not there anymore and I'm not going to overstate what I know. But when I worked there, the idea that one person "owned" the matching engine would have gotten a laugh. It was built and maintained by a whole team. The methods were the kind of thing you'd learn in a graduate computer-vision course or read in a paper. The code was on shared drives the whole team could reach. If there's a real secret in there, it's the fine-tuning and the data, and that's not something that lives in any one engineer's head as a single thing you could hand to a competitor.

I don't know what Priya did in her final days — I'd been gone a long time by then. I can only speak to the covenant and the culture around it, and to how the company treated the same agreement when I left. I'm willing to say all of this under oath.

/s/ Naomi Vandermeer

BUSINESS EXHIBIT · ENGAGEMENT LETTER

Engagement Letter

ELLINGBOE & RAVNDAL LLP 310 Riverwalk Plaza, Suite 400, Norsholm, Meridian 55901

Date: February 17, 2026

Ms. Priya Iyer 27 Riverbend Terrace, Apt. 3B Beacon, New York 12508

Re: Representation in Hudson Valley Biometrics, Inc. v. Iyer — Opposition to Preliminary Injunction

Dear Ms. Iyer:

Thank you for retaining Ellingboe & Ravndal LLP. This letter sets out the terms of our engagement.

Scope of representation. We will represent you in connection with the motion for a preliminary injunction brought by Hudson Valley Biometrics, Inc. by Order to Show Cause in the Commercial Division of the Supreme Court of the State of New York, Dutchess County, seeking to enforce the non-competition, non-solicitation, and confidentiality covenants in your 2019 employment agreement, and in related pre-answer proceedings. This engagement covers our work opposing the injunction, advising you on your rights and risks, and, if appropriate, pursuing a negotiated resolution. It does not include any appeal or any separate trade-secret damages trial, which would require a further agreement. Because the matter is pending in New York, we will associate local counsel as the court's rules require; any such arrangement will be disclosed to you.

Fees. Our fees are charged on an hourly basis. Amara Ellingboe bills at $250 per hour and Marcus Ravndal bills at $225 per hour. Paralegal time, when used, is billed at $120 per hour. We record time in tenth-of-an-hour increments. The hourly rate governing this engagement is $250 for senior-attorney time.

Costs. You are responsible for out-of-pocket costs we advance on your behalf, such as filing fees, transcript fees, and service costs. We will identify these separately on your statements.

Billing. We send statements periodically, typically monthly. Statements are due within thirty days of receipt. Please tell us promptly if you have questions about any charge.

No guarantee of outcome. We will advocate vigorously for you, but we cannot and do not guarantee any particular result. Injunction practice moves quickly and outcomes depend on the facts and on the court's discretion.

Your candor. Our advice is only as good as the information you give us. Everything you tell us in the course of this representation is protected by the attorney-client privilege. We ask that you be complete and candid with us, including about facts you find embarrassing.

If these terms are acceptable, please sign below and return one copy.

Sincerely,

/s/ Marcus Ravndal Ellingboe & Ravndal LLP

AGREED AND ACCEPTED:

/s/ Priya Iyer Date: February 17, 2026

BUSINESS EXHIBIT · INTAKE & CONFLICTS

Intake (2026-02-15). Former employer Hudson Valley Biometrics, Inc. moved by Order to Show Cause for a preliminary injunction to enforce a two-year, geographically unlimited non-compete and trade-secret covenants against Ms. Iyer, alleging she will inevitably disclose its 'matching engine' algorithm if she joins competitor Sightline Perception. Firm engaged to oppose the injunction. Client insists she took no files; employer's own forensic review found no exfiltration.

Conflicts check (2026-02-16) — CLEAR. No prior or current representation of Hudson Valley Biometrics, Sightline Perception, or Halbrook Renner & Iselin. No firm relationship with Mr. Steinhauer. No positional conflict identified in restrictive-covenant defense.

BUSINESS EXHIBIT · BILLING STATEMENT (17 TIME ENTRIES)

BUSINESS EXHIBIT · INVOICES