AI Integration
I take LLM features from prototype to production. Retrieval pipelines, agent workflows, prompts that hold up with real users, and evals to keep them honest.
- RAG & search — embeddings, hybrid retrieval, reranking, citations
- Agents & tool use — multi-step workflows, function calling, guardrails
- Evals & observability — offline benchmarks, online tracing, regression suites
- Fine-tuning & distillation — when prompts hit their ceiling