Building RAG-Powered Product Experiences

This hands-on workshop introduces the core concepts behind Retrieval-Augmented Generation (RAG) and shows how they come together in a working application. We'll start by breaking down the RAG pipeline step by step: embeddings, semantic search, chunking, vector databases, retrieval, ranking, context construction, and generation. Through working code, we'll explore how each stage affects the quality and reliability of the final response, along with common failure modes and practical considerations when building RAG systems. We'll then add a conversational interface on top of the retrieval pipeline to show how RAG can be used inside real product experiences rather than as a standalone search demo. The workshop is approximately 80% hands-on and 20% theory, with two parallel implementation tracks: Python: Sentence Transformers + ChromaDB + PydanticAI TypeScript: Sentence Transformers + ChromaDB + Vercel AI SDK By the end of the workshop, students will understand the major components of a RAG system, the tradeoffs involved in designing one, and the engineering practices needed to build more reliable AI-powered search and conversational product features.