AI Engineering
AI that belongs in the product, not the pitch deck.
I design and ship LLM features, agents, and knowledge systems for founders who need them in production — evaluated, observable, and wired into the rest of the stack.
What this covers
- AI agents and agentic workflows
- LLM-powered applications
- RAG and knowledge systems
- AI automation
- Document intelligence
- AI integrations and product features
- Evaluation, reliability, and AI infrastructure
How I approach AI work
Product first
The model is not the product. We start from the user job, the data you actually have, and the failure mode you cannot afford.
Reliability next
Evaluation, fallbacks, and traces so an agentic workflow can be operated — not only demoed once.