The role :

You'll take over day-to-day backend ownership from our founding engineer, who is shifting to product. The core of the job is building: new API servers and data models across our platform some already in flight, more planned on Python and Azure.

You'll also maintain and extend our production multi-agent system (Python/LangGraph fronting a healthcare data warehouse): it's largely built, and needs incremental improvement, monitoring, and steady hands. You'll carry the backend through a major go-live this fall, working directly with the founders and our external web-app development partner.

What you'll do :

- Design and build new API services and data models greenfield development is the bulk of the role, picking up in-flight work and starting planned services from scratch.

- Maintain and extend the production LangGraph multi-agent backend : incremental capabilities, tool integrations, monitoring, and reliability.

- Design and build multi-tenant access control : per-organization data isolation, credential architecture, JWT/OIDC auth flows (Auth0).

- Run the backend's Azure footprint day to day : managed identities, Key Vault, CI/CD pipelines, environment promotion (dev - UAT - prod).

- Handle healthcare data responsibly PHI-aware design is part of the job.

- Work async with a founder-led team and an external dev shop; write decisions down.

What we're looking for :

- 5 - 8 years of backend engineering with deep, current Python.

- A track record of designing and shipping API services and data models from scratch not just extending existing systems.

- You've worked with LLM agent systems in production LangGraph preferred, or comparable frameworks with real users behind them.

- Production Azure experience : identity, secrets, app hosting, pipelines.

- Strong SQL (SQL Server and/or Postgres) and multi-tenant data-access design.

- Auth fundamentals : OAuth2/OIDC, JWT.

- Self-directed you'll often be the only engineer on the problem.

- Bonus : US healthcare data (claims, CMS programs), HIPAA experience, Terraform.