Gen AI Engineer

THE ROLE

You'll build production generative AI systems inside the operational layer of healthcare — the workflows underneath clinical care that decide where time, margin, and trust actually go. This is not a research role. This is shipping AI that runs against real data, in real EHR-adjacent environments, with real compliance gates. Your work will live next to credentialing systems, revenue cycle pipelines, decision support tools, and clinician-facing interfaces — and it has to keep working after we hand it back to the client.

WHAT YOU'LL DO

• Design and ship LLM-powered features inside healthcare operations workflows — credentialing, intake, documentation, decision support, automation
• Build retrieval pipelines against clinical content, payer policies, regulatory documents, and internal subject-matter material
• Own evaluation: build the test sets, the rubrics, the monitoring that proves the model is doing what we said it would
• Partner with clinicians and operators directly — your designs will be reviewed by the people who use the systems
• Work inside compliance constraints: HIPAA, SOC 2, AI governance frameworks. PHI-aware engineering from day one
• Hand off documentation, runbooks, and transferable IP at engagement closeout — we build for enablement, not lock-in

WHAT WE LOOK FOR

• Production experience with LLMs in regulated or sensitive-data environments — healthcare, finance, legal, defense
• Strong Python. Comfortable with both modern AI tooling (LangChain, vector databases, evals frameworks) and the boring plumbing (auth, queues, observability)
• You've shipped something a domain expert depends on — not just a demo
• Bias toward systems thinking over framework loyalty. You'd rather build the right pipeline than the trendy one
• Comfort with ambiguity. The right answer is often the third one we tried, after the first two clarified what the problem actually was
• Bonus: clinical informatics exposure, HL7/FHIR, prior consulting or embedded work

HOW TO APPLY

Send a résumé and a short note (3–5 sentences is plenty) about a problem you've solved that sounded easy on paper and turned out to be hard.