Python Developer

About NB1 Health

NB1 Health is a health-testing platform that gets clinical-grade testing into people's hands at home. We ship test kits, process them through partner labs, and turn results into clear, actionable health insights for users across Europe and the Middle East (Austria, Romania, Netherlands, Italy, UK, Switzerland and Dubai). Our engineering team is based in Bucharest and builds the platform that runs it all — from kit logistics and lab integrations to the AI features that help users understand their results.

The role

We're hiring a Senior Python Developer who builds production AI systems — not demos. You'll design and ship FastAPI services backed by PostgreSQL, wire them to LLM providers (OpenAI, Anthropic, Google), and build multi-step agent workflows that hold up when the model is having a bad day. A real slice of the job is making AI behave in production: structured outputs, tool calls, retries, cost budgets, latency tails, evals and traces — the unglamorous half of GenAI that separates a cool demo from a feature customers depend on every day. In healthcare, output quality isn't optional — so if you care about getting AI right when it matters, you'll feel at home.

What you'll do

• Design and ship Python/FastAPI services — REST and async streaming endpoints, Pydantic schemas, clean OpenAPI specs, proper error semantics.
• Model and query PostgreSQL — schema design, indexing, migrations (Alembic), JSONB and pgvector where they earn their keep.
• Integrate LLMs directly via provider SDKs (OpenAI, Anthropic, Google) — function calling, structured outputs, streaming, token & cost budgeting.
• Build agent workflows with the right tool for the job (LangGraph, CrewAI, or plain Python when a framework is overkill) — handling tool calls, retries, fallbacks and human-in-the-loop.
• Own the eval + observability layer — prompt regression suites, tracing (Langfuse / LangSmith / Phoenix), cost & latency dashboards, drift detection.
• Own production reliability — timeouts, backoff, circuit breakers, provider failover, prompt caching, rate-limit handling.
• Write tests that catch real regressions, and mentor mid-level engineers.

What we're looking for — Must have

• 4–7 years building production Python services.
• FastAPI (or equivalent async framework) in production — async/await, dependency injection, streaming.
• Strong PostgreSQL — schema design, indexing, migrations; you can read an EXPLAIN ANALYZE and act on it.
• Hands-on LLM API integration with at least one of OpenAI / Anthropic / Google — function calling, structured outputs, prompt tuning past "good enough."
• Solid fundamentals — Git, code review, CI/CD, Docker, observability (logs, metrics, traces).
• Testing discipline — pytest, fixtures, golden-set evals for LLM behaviour.
• Working English (tickets and docs in English).
• Genuine interest in the AI ecosystem — you can name two recent model releases without Googling.

Nice to have

• Production agent-framework experience (LangGraph, CrewAI, AutoGen, LlamaIndex, or hand-rolled).
• LLM eval & observability tooling (Langfuse, LangSmith, Phoenix, Promptfoo, DeepEval).
• Cost & latency optimization — prompt caching, model routing, batching, semantic caching.
• Vector search / RAG — pgvector, Qdrant, Weaviate, Pinecone.
• Domain experience in healthcare, fintech, legal, or another high-stakes field.

Our stack

Python 3.11+ · FastAPI + Pydantic v2 · PostgreSQL (+ pgvector) · Alembic · LLM providers [OpenAI / Anthropic / Google] · Agent framework [LangGraph / CrewAI / hand-rolled] · Eval & tracing [Langfuse / LangSmith / Phoenix] · Docker · CI/CD [GitHub Actions / GitLab CI] · Observability [OpenTelemetry / Grafana / Datadog / Sentry]. We're not chasing tools — every choice earns its place.

What we offer

• Senior-level salary competitive with the Bucharest market range between 3000 - 5000 euro net, reviewed annually on impact.
• Private medical insurance at Regina Maria
• Training & conference budget (PyCon EU, EuroPython, Anthropic/OpenAI dev days).
• LLM API budget for prototyping and learning.
• Modern hardware — top-spec MacBook Pro / Linux workstation, dual 4K monitors.
• A team that takes AI craft seriously: we read papers, evaluate what we ship, and don't believe the hype just because it has a logo.

NB1 Health is an equal opportunity employer. We hire on craft, curiosity and how you think.