Most AI products don’t “train a custom model” on your private data. They use retrieval — a system that finds the exact information an LLM needs in the moment, so it can answer with accuracy instead of hallucinations.
Today’s workshop demystifies real-world RAG (Retrieval-Augmented Generation) and shows you how developers actually connect LLMs to personal notes, emails, documents, or company data — without expensive fine-tuning or fragile hacks.
You’ll explore both sides of retrieval: fast keyword search with BM25, semantic search with embeddings, and how modern systems blend the two using rank fusion. Then you’ll take it further by rewriting user queries and reranking results to dramatically improve retrieval quality.
You will learn how to:
- Understand RAG fundamentals and connect LLMs to private data safely
- Master keyword search with BM25 and semantic search with embeddings
- Combine multiple retrieval methods using reciprocal rank fusion
- Optimize retrieval quality through query rewriting and intelligent reranking
- Build a retrieval-augmented chat endpoint that answers from your data, not model guesses
By the end of this workshop, you’ll know how to build retrieval-augmented AI systems that are fast, accurate, up-to-date, and actually capable of working with private data — the same techniques behind apps like Notion AI, Perplexity, and ChatGPT’s file search. Not just “RAG theory,” but hands-on, working retrieval pipelines you can plug into your own apps.