genai-stack

Why we included this project

This stack is a complete, runnable example of a retrieval-augmented generation app built from real parts instead of tutorial snippets. Docker Compose ties together LangChain, Neo4j, and Ollama, so you can start a working system on your own machine and trace how the pieces connect. It includes several small demo apps: a support agent that answers questions with and without RAG, a loader that ingests Stack Overflow data into a knowledge graph with vector embeddings, a PDF reader that chunks and embeds a local document, and a streaming API with a Svelte front end. Since you can switch between local Ollama models and cloud LLMs with environment variables, it's a good base for prototyping a knowledge-graph-backed Q&A system before you commit to a production setup.