RAG Pipelines Mastery Path¶

Mastery Path

BOOTCAMP

VINOD BHANUJA

A structured, module-wise training program on Retrieval-Augmented Generation (RAG) — from your first embedding to production-grade, master-level RAG systems — with runnable Python code in every module and a hands-on project at the end of each level.

RAG is the technique behind almost every "chat with your documents" product: instead of hoping a language model memorized your data, you retrieve the relevant passages at question time and let the model generate an answer grounded in them. This site teaches that pipeline from first principles, in plain Python — no framework required to understand what's actually happening.

How the program is organized¶

Level	Focus	Modules
Level 1 · Entry	Embeddings, chunking, vector stores, retrieval, grounded generation, a full working pipeline	9 topics + 1 project
Level 2 · Intermediate	BM25 & hybrid search, reranking, query rewriting, table RAG, LangChain & LlamaIndex	9 topics + 1 project
Level 3 · Advanced	Agentic RAG, multi-hop retrieval, GraphRAG, production vector DBs, eval at scale	9 topics + 1 project
Level 4 · Master	Enterprise architecture, multi-tenancy, cost/latency optimization, RAG security	9 topics + 1 capstone

What you need¶

Python 3.10+ and pip. Level 1 uses two free, local, no-API-key libraries: sentence-transformers for embeddings and chromadb for the vector store.
The generation step (turning retrieved text into an answer) is shown with the Anthropic API (pip install anthropic), which requires an API key. Every other part of the pipeline runs fully offline, and any chat-completions API works the same way — the lessons flag exactly where a key is needed.

How to use this site¶

Work through each level in order — later modules assume earlier ones.
Every topic page has runnable code — copy it into a local .py file and run it. Code that needs an API key says so explicitly.
Each level ends with a project that combines everything learned in that level.
Use the search bar (top of the page) to jump straight to a topic.

Start here → Level 1 · Entry

RAG sits between machine learning and LLM application development. Two sister sites cover the neighboring ground: