Consulting
I build the unglamorous engineering that makes AI reliable outside a demo.
RAG reliability
Your RAG works in a demo but returns wrong or empty answers in production. I find where it actually fails — retrieval, chunking, or synthesis — and fix it.
Retrieval quality
Chunking, hybrid search, reranking, and query understanding tuned to your data, not a tutorial — measured, not guessed.
Agentic search
Tool-using LLM systems that decompose, route, retrieve, and self-check — with the latency and cost tradeoffs made explicit.
How I work
Short, focused engagements: a scoping call, a clear diagnosis, and hands-on fixes you can ship. I work directly in your codebase and leave you with the understanding, not just the patch.
Work with me
I take on a small number of consulting projects around RAG, search, and LLM systems. Tell me what you’re building and where it’s stuck.