{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/d06f324022714527a00d1ff94a460e68\" frameborder=\"0\" width=\"1796\" height=\"1347\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1347,"width":1796,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1347,"thumbnail_width":1796,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/d06f324022714527a00d1ff94a460e68-0ce6c8fa8ba88693.gif","duration":529.387,"title":"Medical RAG for FDA and WHO QA","description":"This Loom presents the Medical RAG project, a retrieval augmented generation system for answering questions about regulatory and responsible AI based on FDA and World Health Organization documentation. The presenter emphasizes that it is not a medical diagnosis or treatment tool, but a domain Q and A system, with ingestion, chunking with metadata, embedding via AWS Titan, and storage in a pgvector database. At query time, questions are embedded, similar documents are retrieved, and reasoning is handled using AWS Bedrock with an OPPOS 4.6 model, exposed through a FastAPI and MCP server with a simple UI. The demo shows example questions, estimated response time, and answers with citations and source references."}