I build AI systems and put them into production. Not slideware — a working, deployed system I own from discovery through go-live and hand off clean. Fifteen-plus years shipping client-facing software, now focused on LLM/RAG/agent builds.
What I do
- Discovery → architecture. Turn an ambiguous AI idea into a concrete, buildable spec: data flow, models, integration points, success metrics.
- Build. Implement it in Python / TypeScript / Node on AWS — APIs (FastAPI), data pipelines, front ends (React / Streamlit), and the LLM layer (OpenAI / OpenRouter / Anthropic, LangChain).
- Deploy. Ship a live system with the operational pieces that keep it running.
- Prove it. Evals and real-usage validation so we know it works, with measurable outcomes.
How we can work together
- Scoped paid pilot — a 2–4 week fixed-price deliverable (e.g., a working RAG prototype over your docs, with evals) so you see the work before committing headcount.
- Full build — end-to-end delivery of a production AI feature or system.
- Fractional / embedded engineering — senior AI direction plus hands-on delivery for a team that’s scaling.
Why me
I own the customer outcome, not just the code — the forward-deployed profile. You get one operator who can talk to your users, make the architecture calls, write the code, and stand it up in production.