Skip to content

Case Studies

Real projects showing how I think about AI feature development. The focus is on failure handling, user experience, and what it takes to go from demo to production.

  • RAG Support Assistant

    A customer support RAG pipeline built around what happens when the AI can't answer confidently. Five-category query classification, two-stage retrieval with confidence gates, self-critique, and designed failure paths for every edge case. Built with OpenAI, ChromaDB, FastAPI, and a custom chat widget.