Product Engineer

Product Engineer — Regulatory AI Platforms
Engagement: Part-time independent contractor — ~25–35 hours per week, flexible scheduling
Location: Fully remote (United States)
Term: Initial 6-month engagement with a defined path to an expanded or full-time role as we scale
Compensation: Hourly, commensurate with experience
About Us
LexReg is a regulatory intelligence platform for banks, credit unions, and fintechs. We offer the LexRegPulse and LexRegulator platforms. LexRegPulse monitors more than 130 federal and state regulatory and industry sources overnight and publishes a synthesized intelligence brief by 6:45 AM ET. Underneath the brief sits the thing that actually matters: a structured, versioned corpus of U.S. banking regulation, broken down into individual, separately citable obligations — the corpus that powers LexRegulator, our charter-scoped AI compliance agent.
Our product promise is that every claim we make traces to a real obligation, every obligation traces to a pinned primary source, and every source is being watched for change. Nothing is paraphrased. Nothing is asserted without a citation a customer can click. Only signal, no noise. This is the bleeding edge of legal and compliance utilization of Agentic AI and the future of the practice.
That promise is only as good as the systems that enforce it.
The Role
You will work with our team and help build products, end to end, across three areas:
Partner platforms. Products that other companies’ engineers integrate against. You take these from a conversation with the business to production.
The corpus and the checks around it. Collectors for sources that publish a PDF with no stable URL and no change notification. The schema and its versioning: supersession, amendments, effective dates, and applicability by charter type and asset threshold.
The tooling and the alarms. The review console, and the per-source health, extraction-yield, and alerts.
How We Work
Agents propose, humans dispose. Automation drafts; it never commits. The commit moment is always a human click. Your job is to make that click well-informed, fast, and reversible —never to route around it.
Work arrives in a queue with a recommendation attached. If we are relying on anyone to remember to check something, we have built it wrong.
Fail loud, never fail quiet. Silent degradation is the only unacceptable outcome.
Deterministic beats probabilistic. If a check can be a string comparison, a schema constraint, or an HTTP status code, it is not an LLM judgment.
The same standard applies to our own code. We have shipped a quality bar no score could reach, and a verdict computed from a field that was never loaded. Both passed review. Finding that class of defect is the job, not an interruption to it.
Who You Are
We hire for the bar, not the résumé.
A current undergraduate or graduate student in engineering, computer science or a related field, or a recent graduate, able to commit 25–35 hours per week.
Working TypeScript and SQL, and a real project you can show — shipped, public, or from an internship or research — that ingests or parses data from a source that did not cooperate.
Or
Two or more years building data-intensive systems in production and keeping them honest. Fluent in TypeScript and SQL; comfortable in Postgres and Prisma. You have designed a schema you later had to migrate, and you have opinions about it now.
You have parsed hostile inputs: inconsistent HTML, PDFs with no text layer, tables that are images, and pagination that changes shape.
You have shipped something that works with an LLM, and you are appropriately skeptical of it. You have been burned by a model that was fluent and wrong; you built a check instead of a better prompt.
Required
Familiarity with AI platforms and using AI to enable efficient development.
You work alongside agentic coding tools and treat their output as a draft written in a confident voice. You can tell a plausible difference from a correct one. When the build is green, your instinct is to verify the claim against production rather than merge.
No ego about escalating.
Intellectual curiosity and honesty under commercial pressure. You will sometimes be the person saying a pipeline is not good enough to ship.
Helpful, Not Required
Bank or credit union compliance, examination, or supervisory correspondence.
Send a résumé to: Emanuel Slater at: emanuel@lexregpulse.com.
If you want to show us something — problem solving work you are proud of — feel free to share it, and tell us what broke and what you did about it.
LexRegPulse is an equal opportunity employer. We evaluate every applicant on the work.

Work arrangement
Yes