RAG Systems

Design and build Retrieval-Augmented Generation systems to query custom data

Welcome to the Retrieval-Augmented Generation (RAG) module. Learn how to connect LLMs to external, private knowledge sources to ground model answers in facts and prevent hallucinations.

💻 Module Practice Notebooks

Master all the concepts from this module with hands-on practice:

• Practice in VS Code: Open the notebook in your local editor. Requires a local .env file containing your API keys.
• Practice in Google Colab: Open the notebook directly in Colab. Setup cells are included to install packages and request API keys.

💻 VS Code | 🚀 Colab | 📥 Download Notebook

🗺️ Module Overview

Build end-to-end question answering systems over proprietary data:

📚 Module Curriculum & Roadmap

Explore the sections in this module using the cards below:

1. Introduction to RAG

Learn why RAG is required, how it differs from model fine-tuning, the 3 pillars of RAG pipelines, and lexical vs. semantic search comparisons.

2. Document Ingestion & Chunking

Understand document loaders (Text, PDF, HTML), chunk size and overlap constraints, and explore character, recursive, token, and semantic chunking.

3. Vector Databases & Retrieval

Evaluate Cosine vs. L2 distance metrics. Setup ChromaDB, generate embeddings, and build a working similarity retrieval search engine.

4. Generation, Advanced RAG & Eval

Build RAG synthesis chains with LCEL. Compare Naive, Advanced, and Agentic RAG, and analyze context precision, recall, and faithfulness metrics.

🛠️ Practical Capstone Project

”Chat with your PDF” Application

Build a Python-based command-line tool (or Streamlit UI) that:

1. Allows a user to upload a custom PDF report.
2. Automatically parses, chunks, embeds, and loads the content into a local vector database.
3. Takes user questions, retrieves the top-k relevant segments, and passes them to the LLM.
4. Returns the model’s answer along with the exact source pages/chunks used for verification.