Design and ship AI systems that run in production: retrieval pipelines, tool-using agents, evaluation harnesses, and the glue that makes models useful inside real operations.
We partner with Anthropic and ship Claude-backed systems. Novelty without reliability is not the job.
02
Requirements
01Four or more years software engineering with at least two years in applied AI/LLM systems
02Hands-on with RAG, prompt/tool design, and Python or TypeScript services
03Experience evaluating quality, cost, and latency — not only demos
03
Nice to have
+LangChain, LlamaIndex, or custom agent runtimes
+MCP (Model Context Protocol) or multi-agent orchestration
Hiring
What happens next
01
We read it
A person reads every application. No keyword filter.
02
Reply
Within 5 working days if there is a fit.
03
Conversation
90 minutes with the people you would work with.
04
Paid trial
A real, scoped problem. Paid at contract rate.
Application
Send your work. Skip the theatre.
Resume, a short note, and anything that shows how you think. Specifics beat adjectives.
Show something you built, not something you managed.
Scale, constraints, and trade-offs land better than adjectives.