Full-Stack Engineer

About Lab651

Lab651 is a software engineering firm for regulated industries, based in Saint Paul, Minnesota. We build custom software for medical device, healthcare, insurance and financial services companies, where getting it wrong is expensive and the engineering has to hold up to scrutiny. Enterprises and funded startups outside regulated work hire us for the same rigor.

Our approach: AI that moves fast, engineering that holds the line. AI accelerates the work. Our engineers own every line of it. Every project runs on the Lab651 Proven Project Process™, with AI agents in every phase and the spec at the center. When a project is regulated, we deliver against FDA 21 CFR Part 11, IEC 62304, HIPAA or SOX.

We are a small team. Everyone works directly with clients, and everyone ships.

The role

We are hiring a full-stack engineer who uses AI to build faster and uses engineering discipline to build right. You will own features end to end, from the written spec through implementation, tests and release, on products where a defect can mean a failed audit or a patient at risk.

This is not a ticket-taker role. You will work alongside the product owner, UI/UX, QA and the project manager on each project. You will sit in client meetings, ask why before how, push back when a requirement does not hold up, and be trusted to run a workstream on your own. You will also be expected to make the people around you better: pair, review, share what works.

What you’ll do

Build and ship full-stack features: web front ends, APIs, cloud services and the data behind them.
Work across the whole process: shape requirements in Discovery, turn the spec into tasks and test plans at engineering kickoff, build in agile sprints, and support the client through user acceptance.
Work to spec. Turn requirements into REQ-numbered specifications, implement against them, write tests per requirement, and keep the traceability matrix generated from the work rather than maintained by hand.
Pair with AI agents every day, using Claude Code, Copilot, Codex or similar: research and requirement drafts, planning, implementation, test generation, first-pass code review and release evidence. Review every line the AI produces. You are accountable for it.
Lead client conversations. Run discovery sessions, present options and tradeoffs, give status without being asked, and say no when the scope or timeline is wrong.
Own a workstream independently. Plan it, estimate it, deliver it, and raise risks early.
Deliver inside regulated constraints: documented requirements, traceable design, verifiable behavior, audit-ready records.
Review teammates’ work, pair when it helps, and write down what you learn so the team gets faster together.

What we’re looking for

5+ years shipping production software across the stack: front end, back end, data and cloud. We are polyglot by necessity. The more languages and frameworks you can work in, the better. Typical stacks include TypeScript and React, Python, C# and .NET, with Azure, AWS or GCP behind them.
Daily, hands-on use of AI coding tools, and a clear point of view on where they help and where they fail. You can show us a workflow, not just a prompt.
Experience building to written specifications: requirements documents, acceptance criteria, test plans. You are comfortable being measured against what the spec says, and you flag it when the spec is wrong.
Product-minded. You ask who the user is and what problem this solves before you open the editor.
Comfort in front of customers. You can run a meeting, explain a technical decision to a non-engineer, and take a hard question without getting defensive.
Self-direction. You can take a loosely defined outcome, turn it into a plan, and deliver it without being managed day to day.
Collaboration as a default. Small team, no silos. You share context, review generously, and would rather ask than guess.
Clear writing. Specs, PRs, status updates and client emails are part of the job.

Nice to have

Work in a regulated environment: medical device software (IEC 62304, FDA 21 CFR Part 11), HIPAA or SOX.
Spec-driven development with tools like Spec Kit, or systems engineering practice (INCOSE, MBSE).
Consulting or agency experience, with several clients at once.
Mobile (iOS, Android, React Native) or embedded and IoT exposure. Much of our work runs device to cloud.
Building with LLM APIs: RAG, agents, evals, prompt and context engineering.
Security practice: threat modeling, passing a customer security review.

How we work

Every project follows the Lab651 Proven Project Process™: Discovery, Project Kickoff, Engineering Kickoff, Agile Sprint Iterations and User Acceptance. The spec is the single source of truth. It is drafted in Discovery, refined at kickoff, built and verified against in every sprint, and traced at acceptance. See the process.

No spec, no build. Inside every sprint we use Spec-Driven Development: Specify, Plan, Tasks, Build, Verify. Every feature starts with a written requirement, and every commit traces back to one.

AI agents work in every phase, drafting and checking. Humans decide, approve and own the risk. We sort work by risk: routine drafts get a light review, requirements and test generation get domain-expert validation, and safety-critical logic or regulatory artifacts get full formal review. You will work inside that ladder and help refine it.

Small team, extreme transparency. You will know what every client is paying for and why. Clients see our plans, our estimates and our progress. No surprises in either direction.

We teach as much as we build. Our team runs AI-enablement workshops for Fortune 500 engineering groups and teaches graduate AI courses. If you like explaining things, you will fit.

Details and how to apply

Item	Detail
Location	Remote is fine. Hybrid is welcome: our office is in Vandalia Tower in Saint Paul, MN, next door to Barrel Theory Beer Company.
Type	Contract to hire. 1099 to start.
Compensation	Hourly, based on experience.

To apply, email connect@lab651.com with a short note on a project you shipped using AI-assisted engineering: what the tools did well, where they failed, and how you caught it. A link to code, a PR or a write-up is worth more than a resume. We reply within one business day.