We hire engineers, not prompt engineers. You'll spend roughly half your time writing TypeScript and Python around LLM APIs, a quarter doing customer engineering directly (you'll meet the people whose workflow you're replacing), and the rest making sure what you ship keeps working in production.
What you'll do
- Own one or two customer engagements end-to-end — from kickoff conversations through production handoff.
- Design agent architectures using OpenAI, Anthropic, Gemini, LangGraph and the right tool for the job.
- Build the eval harnesses, retrievers, function-calling layers, and observability that turn a demo into a production system.
- Pair with our customers' engineers — they should leave the engagement with a system they can run themselves.
- Write postmortems when things break and make sure they don't break the same way twice.
What we look for
- 5+ years shipping production software in TypeScript or Python.
- 1+ year building with LLM APIs in production — not just prompt-and-pray demos.
- Comfort with retrieval, function calling, multi-turn state, and the failure modes that come with them.
- Strong written communication. We work async with customers across time zones.
- You enjoy talking to customers and translating fuzzy business problems into concrete eval criteria.
Nice to have
- Experience with LiveKit, Twilio, or production voice systems.
- Background in a regulated industry (healthcare, fintech).
- Open-source contributions to LangChain, LangGraph, LlamaIndex, or similar.