Primary AI is a clinical decision support tool built by a physician. It answers clinicians' questions with responses grounded in medical literature, and every citation is machine-verified to actually support its claim. The product is live with quick-growing early usage.

The honest pitch: the system was built quickly by one founder working with AI coding agents. That shipped a real product: a corpus ingestion pipeline, hundreds of edge functions, scheduled jobs, and a citation verification model in training. What it needs now is an experienced engineer keeping production trustworthy. Bugs here fail silently, mistagging data or quietly halting ingestion until an audit catches them. Your job is to make the system tell us before the data does.

The work:

- Own reliability of the corpus ingestion pipeline: correctness, provenance, monitoring
- Harden CI and deploys, add observability so silent failures become loud ones




- Review changes produced by AI coding agents before they reach production
- Help productionize the citation verification model training loop

Stack:

TypeScript (Deno) on Supabase, Postgres, GitHub Actions, Python for model training. You don't need all of it on day one, just a track record of owning unfamiliar production systems.

You're a fit if:

- You've run production systems end to end and care deeply about data integrity
- Strong SQL/Postgres, comfortable in TypeScript and Python
- Rigorous about verification: "it deployed" and "it works" are different claims to you
- Available for a consistent weekly commitment, with room to grow into a founding engineer role if it's a fit

📌 Contract Engineer, Primary AI (Clinical Decision Support) (Laval)
🏢 HealthTech Investors
📍 Laval