Participated in public demo 2510 with CWYT

The problem [00:40] Family history and oral stories tend to live scattered across documents, memories, and relatives rather than in one place a person can easily explore or ask questions of. What they built [00:40] CWYT ("Conversation With Your Thoughts") is a conversational RAG platform: a vector database retrieves relevant family stories for a query, and an LLM generates a grounded narrative answer, citing how many stories were retrieved. A companion Story Review interface lets him curate, edit, and publish the underlying stories that feed the system, with passwordless magic-link login behind an approval-gated access model. What happened [05:10] Building the system wasn't the hardest part, making it reliable was. Wayne built evaluation datasets and LLM-judge diagnostics scoring retrieval quality across relevance, groundedness, and completeness, and used them to uncover and fix a broken vector search that had been silently undermining result quality, a failure mode traditional debugging would have missed. The ask [07:00] Wayne closes open to work, looking for a Technical Program Manager role focused on AI evaluation.