Founding AI Engineer

The role: You're our founding AI engineer — the person who owns the intelligence layer of the product. You'll design and ship the agentic systems that read our healthcare data, reason over it, call tools, and put decisions in front of finance and operations leaders: LLM orchestration, retrieval, tool-use contracts, evals, guardrails, and the inference stack. You'll also own the data foundation that makes the AI trustworthy — the messy, high-stakes healthcare data (claims, clinical, RCM) the models run on. No inherited stack, no playbook. A direct line to the CTO and your fingerprints on every decision that matters.

Who you are: a bar on talent and trajectory, not a checklist of tools

• You've shipped real AI in production — LLM-powered, agentic, tool-using systems serving real users. You've hit the hard parts (evals, latency, reliability, hallucination, cost) and shipped anyway.
• You've been a founding or early engineer at a startup — or an exceptional engineer at a top company itching to build from zero. You move fast and own outcomes end to end.
• You're fluent with data, and you've worked with messy, regulated, high-stakes data. Healthcare claims (837/835/834), clinical, or RCM experience is a real requirement here — you own the data foundation as well as the AI, and the AI is only as good as the data under it.
• You're US-based and authorized to work in the US without visa sponsorship, now or in the future. What matters less than you'd think
• The exact tools on your résumé. Our stack is modern (Python, modern LLM/agent frameworks, a cloud warehouse, dbt, Postgres), but we'd rather hire a brilliant builder who'll master ours than someone who's memorized a specific one.
• Years of experience as a number. We've watched 5-year builders out-build 15-year veterans. We index on slope, not tenure.
• A traditional enterprise path. If you're happiest tuning a mature system inside a big org, this isn't it — this is a 0-to-1 seat. Why this seat is rare
• "Founding" is usually just a title. Here it's the job — you're the first AI hire, you define how an entire healthcare-AI company uses AI, and you report straight to the CTO.
• The backing is real. AlleyCorp doesn't just write checks; it builds category leaders from zero — MongoDB, Pearl Health, Rogo — and holds its AI to a "real work, not copilots" bar. You get a rocketship's backing with a garage's ownership.
• You build real AI, not a copilot demo. Live customers and production healthcare data from day one — you ship into real workflows instead of waiting on a dataset.
• Real ownership: an equity stake in an early-stage company, behind a backer whose last data company is worth tens of billions. Compensation