Develop RAG application (Retrieval-Augmented Generation) in practice

• 3 days

Outline

Day 1 : Foundations of RAG & Setup

• Introduction to Retrieval-Augmented Generation (RAG)
  • What is RAG?
  • Why is it important?
  • Use cases and real-world applications
  • Comparison of RAG vs. Traditional LLM responses (Prompt engineering)
• Understanding the Components of a RAG System
  • Document Retrieval (Vector Databases, Keyword Search)
  • Embeddings & Similarity Search
  • Large Language Models (LLMs) and their role in RAG
• Building a Simple RAG Pipeline
  • Data ingestion & preprocessing
    • Structured data
      • Database
    • Unstructured data
      • PDF files
  • Generating embeddings with LLM provider
    • Embeddings
    • Vector database
  • Implementing a basic retrieval system
  • Workshop
    • Creating a basic RAG system with a local document store
    • Use cases
      • Chatbot
      • Question/Answering from knowledge (Database and PDF file)
      • Log analysis
      • Data analysis

Day 2 : Enhancing RAG with Advanced Techniques

• Deep Dive into Embeddings and Vector Search
  • Types of embedding models (OpenAI, Hugging Face and Amazon Bedrock)
  • Understanding vector search algorithms (FAISS, Pinecone, ChromaDB)
  • Workshop
    • Implementing vector search using ChromaDB
    • Use cases
      • Chatbot
      • Question/Answering from knowledge (Database and PDF file)
      • Log analysis
      • Data analysis
• Optimizing Retrieval in RAG
  • What is Chunking in RAG ?
    • Improved Accuracy
    • Enhanced Efficiency
    • Preserved Context
    • Information Access
  • Chunking strategies for better retrieval (Chunking Considerations)
    • Fixed Size Chunking
    • Recursive Chunking
    • Document Based Chunking
    • Semantic Chunking
    • Agentic Chunking
  • Hybrid search
    • Combining keyword-based and vector search
  • Ranking and filtering techniques for retrieved documents
  • Workshop
    • Implementing a hybrid search
    • Use cases
      • Chatbot
      • Question/Answering from knowledge (Database and PDF file)
      • Log analysis
      • Data analysis

Day 3 : Deploying & Scaling RAG Applications

• Integrating RAG with APIs and Web Applications
  • Exposing RAG as a REST API using FastAPI
  • Frontend integration with web apps (Streamlit)
  • Workshop
    • Deploying a simple RAG-based
      • Chatbot
      • Question/Answering from knowledge (Database and PDF file)
      • Log analysis
      • Data analysis
• Scaling and Performance Optimization
  • Caching responses for faster results
  • Distributed search and multi-vector index strategies
  • Handling large-scale document ingestion
  • Workshop
    • Optimizing a RAG pipeline for high performance
    • Use cases