hwdsl2/self-hosted-ai-stack
Deploy a complete self-hosted AI stack with Docker Compose. Includes Ollama, AnythingLLM, LiteLLM, Whisper, Kokoro, Embeddings, Docling, and MCP Gateway. Local-first, private by default, multi-arch (amd64, arm64) with optional NVIDIA GPU acceleration.
GitHub repository with 85 stars and 17 forks.
Language: Shell
Topics: ai, ai-stack, docker, docker-compose, docker-image, embeddings, inference, linux, llm, local-ai