I Built an AI That Understands Any GitHub Repository (100% Local)

Ever joined a new project and spent hours trying to understand the codebase? In this video, we'll build a fully local Codebase Q&A Agent that indexes any GitHub repository and lets you ask architecture questions in plain English. No cloud APIs. No OpenAI. No sending your source code anywhere. Everything runs locally using Python, Ollama, ChromaDB, and Streamlit. 🚀 In this video you'll learn: ✅ How Retrieval-Augmented Generation (RAG) works for source code ✅ Building a local GitHub repository indexer ✅ Chunking source code for better retrieval ✅ Generating embeddings with Ollama ✅ Storing vectors in ChromaDB ✅ Semantic code search ✅ Streaming AI responses ✅ Grounded answers with file & line citations ✅ Software architecture Q&A using local LLMs ✅ Extending the system with reranking, AST chunking, and multi-repo search By the end of this tutorial, you'll have a complete AI application that can explain architecture, trace code flows, and answer questions about any repository—all while keeping your code completely private. 🛠 Tech Stack • Python • Streamlit • Ollama • ChromaDB • RAG • Local LLMs • GitHub ⭐ GitHub Repository https://github.com/thecuriousarchitec... If you enjoy practical AI engineering tutorials covering Agentic AI, RAG, Ollama, MCP, and local LLMs, subscribe for new videos every week. #AI #RAG #Ollama #Python #ChromaDB #GitHub #SoftwareArchitecture #LocalLLM #GenerativeAI #AIEngineer #Streamlit #MachineLearning