RAG for a Medical Company: Technical and Product Challenges AI & Data
summary
Explains building a Retrieval-Augmented Generation (RAG) system for a medical company's drug documentation aimed at physicians, using embedding models, vector search, and LLMs (tools mentioned: Chainlit, Qdrant, DVC, Langsmith). Describes improving retrieval and generation metrics, and reducing LLM hallucination risk by showing source HTML snippets and highlighting key facts in responses.
AI-written summary · about
chapters (11)
- 0:00Medical RAG introduction
- 3:21Speaker introduction
- 4:02What is RAG?
- 8:08First system baseline
- 11:51Deployment and monitoring
- 16:45Key technical improvements
- 21:50Product value evaluation
- 26:23Medical challenges and quotes
- 27:13Evaluation process and iteration
- 28:17Takeaways and conclusion
- 30:11Q&A
related talks
51mDeveloping a Custom RAG System
43mAdvanced Retrieval-Augmented Generation Techniques
53mFrom RAG to Riches: Secrets of Data-Augmented LLMs
29mRAG for a medical company: the technical and product challenges
21mHow RAG is Redefining AI Capabilities
42mTo RAG or not to RAG
46mOptimizing Retrieval-Augmented Generation (RAG) for eCommerce
50mRetrieval Augmented Generation Beyond Toy Demos