Retrieval-Augmented Generation (RAG)

About the course

Most enterprise AI value sits in private knowledge: policies, tickets, contracts, wikis, and product docs. This course teaches you to design retrieval-augmented generation systems that answer from those sources instead of guessing. You will cover chunking, embeddings, vector stores, evaluation, and the operational issues that make RAG fail in production.

Learning outcomes

• Design a RAG architecture for a defined document corpus
• Choose chunking, embedding, and retrieval strategies with trade-offs
• Evaluate answer quality, citation coverage, and retrieval hit-rate
• Handle access control, freshness, and hallucination in grounded Q&A
• Ship a working knowledge assistant as a capstone

Hands-on labs and projects

• Index a mixed document set (PDF + markdown) into a vector store
• Build a cited Q&A flow with LangChain or equivalent
• Write an evaluation set and measure retrieval vs generation errors

Technologies