docker/genai-stack

Langchain + Docker + Neo4j + Ollama

What it solves

It provides a ready-to-use starter kit for developers to build and deploy Generative AI applications quickly. It eliminates the initial setup friction by bundling the necessary infrastructure—including LLMs, vector databases, and application frameworks—into a single deployable stack.

How it works

The project uses Docker Compose to orchestrate a suite of tools. It integrates LLMs (via Ollama for local models or API keys for GPT-4, Claude, and Google), a Neo4j database for both vector embeddings and knowledge graph storage, and Python-based applications. It supports multiple deployment profiles for different operating systems and hardware (e.g., Linux-GPU).

Who it’s for

Developers who want to experiment with RAG (Retrieval-Augmented Generation) and knowledge graphs without manually configuring the underlying database and model infrastructure.

Highlights

  • Multi-App Demo Suite: Includes a support bot, a Stack Overflow data loader, a PDF reader, and a standalone HTTP API.
  • Hybrid RAG: Demonstrates the difference between pure LLM responses and responses enhanced by both vector and knowledge graph context.
  • Flexible LLM Support: Compatible with local models via Ollama as well as cloud providers like OpenAI, AWS, and Google.
  • Integrated Tooling: Includes a Svelte-based front-end and supports LangChain tracing for debugging.

Related

  • Project
    n8n-io/self-hosted-ai-starter-kit

    A Docker Compose template that bundles n8n, Ollama, and Qdrant to quickly set up a self-hosted, local AI development environment for building AI agents and workflows.

  • Project
    kossakovsky/selfhost-ai

    A Docker Compose-based platform that deploys over 30 AI and automation tools, including local LLMs, agent frameworks, and vector databases, with a single command.

  • Project
    awslabs/graphrag-toolkit

    A collection of Python tools for building graph-enhanced Generative AI applications using hierarchical lexical graphs and structured knowledge graphs.

  • Project
    neo4j/neo4j-graphrag-python

    The official Neo4j Python package for building GraphRAG applications, enabling the automated construction of knowledge graphs and hybrid retrieval for LLMs.

  • Project
    HKUDS/LightRAG

    A lightweight, graph-based RAG framework that combines knowledge graphs and vector embeddings to provide deep contextual understanding and efficient incremental updates.