Agentic AI - Private Agentic RAG with LangGraph and Ollama

**Private Agentic RAG with LangGraph and Ollama** is an advanced, project-based course that teaches you how to build private, production-ready Retrieval-Augmented Generation (RAG) systems using LangGraph, LangChain, Ollama, ChromaDB, Docling, and Python. This course is designed for developers who want strong control over their data, full privacy, and complete end-to-end workflows using local LLMs. You will learn how to build modern RAG systems, implement advanced retrieval pipelines, add agent workflows, use LangGraph state machines, integrate SQL agents, and run everything on your own machine using Ollama. All projects run 100 percent locally, with no external API cost and no data leaving your system. The entire course is practical. Every concept is explained with step-by-step notebooks, complete Python code, and real examples using SEC financial filings from Amazon, Google, Apple, and Microsoft. What You Will Learn Ollama and Local LLM Setup - Install and configure Ollama for private LLM deployment - Use models like Qwen3, GPT-OSS, Llama 3.2, and nomic-embed - Create custom LLMs with Modelfiles - Use Ollama CLI and REST API for text, chat, and embeddings LangGraph Fundamentals - Build state machines using TypedDict - Create nodes, reducers, and conditional edges - Build multi-step workflows with START/END logic - Visualize execution with diagrams - Understand message accumulation and state merging Complete RAG Systems (from scratch) - Ingest PDFs using Docling with OCR and table extraction - Build page-level chunks for accurate retrieval - Extract metadata from filenames and LLMs - Remove duplicates using SHA-256 hashing - Store documents in ChromaDB with metadata filters Two-Stage Retrieval Pipeline - Build metadata filters from natural language - Generate financial keywords using structured LLM outputs - Use ChromaDB with MMR search - Implement BM25Plus re-ranking for better accuracy - Extract headings and sections for improved ranking Agentic RAG using LangGraph - Build tool-calling agents using the ReAct pattern - Implement document retrieval tools using LangChain - Build agents that call tools multiple times - Add table-based answers with citations - Support multi-turn conversations with memory Corrective RAG (CRAG) - Grade retrieved documents using a Pydantic schema - Detect irrelevant results and rewrite queries - Add web search fallback using DuckDuckGo - Prevent infinite loops with controlled retries - Generate final answers with correct citations MySQL SQL Agent - Build a natural-language SQL agent with LangGraph - Retrieve schema, generate SQL, validate, run, and fix errors - Handle multi-table joins and complex metrics - Automatically correct broken SQL queries - Support explanations and safe database access Financial Document Analysis Project - Work with real SEC filings: 10-K, 10-Q, 8-K - Build a complete RAG system that answers questions like: - “What was Amazon’s revenue in 2023?” - “Compare Google and Apple’s cash flow for 2024” - “Show segment revenue with citations and tables” - Use ChromaDB + BM25 for accurate retrieval - Produce clean, formatted answers with tables and reasoning Who This Course Is For - Developers and engineers who want to build advanced RAG systems - ML practitioners who want full privacy using local LLMs - AI engineers working on LangGraph, LangChain, or agent systems - Backend developers who want to build real GenAI applications - Anyone interested in private, production-grade LLM workflows - This is an advanced-level course. Good LangGraph or Langchain knowledge is required. Why This Course Is Different - The entire course runs locally using Ollama - Zero API cost and complete data privacy - Covers modern RAG techniques: PageRAG, CRAG, Reflexion ideas - Real datasets from top tech companies - Covers LangGraph deeply with real production workflows - Includes SQL agents, financial RAG systems, and multi-step agents - Step-by-step, practical, and code-heavy By the End of This Course You Will Be Able To - Build private, production-ready RAG systems - Deploy and fine-tune local LLMs with Ollama - Build graph-based agents using LangGraph v1 - Create advanced retrieval pipelines using MMR and BM25Plus - Analyze financial documents with precise citations - Build SQL agents for natural language database queries - Handle query rewriting, grading, and web fallback - Build complete agentic RAG applications end-to-end