Skip to content
RAG Repo

Course

Build your own RAG from scratch

Beginner-friendly~1 hour7 lessons

In this course you will build a small but complete RAG system in Python, from raw documents to grounded answers. No frameworks and no magic: just a handful of open tools, so you can see exactly how each piece works. By the end you will have a script that answers questions about your own documents, and a clear mental model you can take to any RAG framework.

Start the course
  1. 1What you will buildThe plan for the course, the RAG pipeline at a glance, and the tools you need installed.
  2. 2Get your dataChoose a small corpus, load it into Python, and tidy the text ready for chunking.
  3. 3Chunk your documentsWhy you split documents into passages, how to pick a chunk size, and a simple chunker in Python.
  4. 4Create embeddingsTurn each chunk into a vector with a free local model, so passages can be compared by meaning.
  5. 5Store and searchBuild a tiny vector store and nearest-neighbour search by hand, with nothing but numpy.
  6. 6Retrieve and generateJoin the pieces by retrieving the best passages for a question and having a language model answer from them.
  7. 7Make it betterThe levers that improve answer quality, how to measure them, and where to go from here.