RAG Lab

Learn to build Retrieval-Augmented Generation systems — from your first 30-line script to production-grade pipelines — running entirely on your own machine.

RAG Lab is a free, open learning portal. No API keys. No cloud bills. Every example is designed to run locally with open-source models via Ollama, so you can experiment as much as you like.

• New to RAG?

  Start with the fundamentals and build your first system in under an hour.

  What is RAG?

• Want the theory?

  Understand embeddings, vector search, chunking, and retrieval from the ground up.

  Foundations

• Just want to build?

  Jump into hands-on tutorials with copy-pasteable code.

  Tutorials

• Going to production?

  Hybrid search, reranking, agentic RAG, evaluation, and scaling.

  Advanced

What is RAG, in one picture

RAG gives a language model access to your data at answer time. Instead of relying only on what the model memorized during training, you retrieve relevant passages from your own documents and augment the prompt with them before generating an answer.

flowchart LR
    Q[User question] --> E[Embed query]
    E --> S[(Vector store)]
    S -->|top-k relevant chunks| P[Build prompt:<br/>context + question]
    P --> L[LLM]
    L --> A[Grounded answer]
    D[Your documents] -.ingest & embed.-> S

The payoff: answers grounded in your content, with fewer hallucinations, citable sources, and no retraining when your data changes — you just update the index.

The learning path

This portal is organized as a curriculum. You can read top-to-bottom, or jump to what you need.

Stage	You'll learn to…
Getting Started	Explain RAG, set up Python + Ollama, and run a local model.
Foundations	Reason about embeddings, vector databases, chunking, and retrieval.
Building Blocks	Assemble an ingestion → index → generation pipeline.
Tutorials	Build a minimal RAG, a ChromaDB-backed app, a chat UI, and PDF Q&A.
Advanced	Add hybrid search, reranking, agents, evaluation, and production hardening.
Tools	Choose between LangChain, LlamaIndex, ChromaDB, FAISS, Qdrant, and more.
Projects	Study and extend full, runnable sample applications.

Why local-first?

Everything here runs offline

We deliberately favor open-source, local tooling — Ollama for LLMs, ChromaDB for vectors, sentence-transformers for embeddings. You learn the mechanics of RAG without a paywall, and what you learn transfers directly to hosted APIs when you need them.

Benefit	Why it matters for learning
No cost	Iterate freely without metering every token.
Privacy	Your documents never leave your machine.
Transparency	You see every component — nothing hidden behind an API.
Portability	The same architecture works with hosted models later.

Ready?

Begin here → What is RAG?

---

About this project

RAG Lab is open source, built by Sitharaj. Found a bug or want to add a tutorial? See the contributing guide. Content is CC BY 4.0; sample code is MIT.

Enjoying RAG Lab?

If this portal helps you learn, consider buying me a coffee ☕ — it keeps the project free and growing. A ⭐ on GitHub helps too!