Retrieval-Augmented Generation (RAG)
This course is for engineers who want a language model to answer questions from their own documents instead of from memory. You will build a RAG pipeline with hybrid search, re-ranking, query rewriting and caching, and measure it well enough to know which stage to fix.
Course Content
Why a language model needs retrieval, how a RAG pipeline is put together, and how to choose between keyword, vector and hybrid search.
How to turn short, vague or conversational user queries into search keys that find the right passage, and how to prove each change helped.
How to re-rank and filter what retrieval returns, cache and remember safely, and measure every stage so you know what to fix next.
Build a RAG chatbot over documents of your choice and prove, with a stratified test set and an ablation table, what each stage contributes.