Explore the directory

A high-performance, open-source large language model library and reference implementation written in pure Python and JAX, aimed at training and fine-tuning models such as Gemma, Llama, DeepSeek, Qwen and Mistral at scale on Google Cloud TPUs and GPUs.

• Models

A highly optimized library of low-level neural network inference operators for ARM, x86, WebAssembly and RISC-V, used as an acceleration backend inside higher-level frameworks such as TensorFlow Lite, PyTorch, ONNX Runtime and MediaPipe rather than directly by model developers.

• Inference

An open-source, cloud-native, one-stop AI platform covering traditional machine learning, deep learning and large-model workflows, from multi-tenant compute scheduling through pipeline orchestration to model deployment and an LLMOps agent layer.

• Data engineering
• Hosting

A terminal client for chatting with Ollama, OpenAI, Anthropic and any other provider supported by pydantic-ai, built as a text-based user interface.

• Inference

Tutorials, demos and example notebooks for building applications on the Milvus vector database, covering retrieval-augmented generation, semantic and hybrid search, recommendation and other unstructured-data search use cases.

• RAG
• Tutorials

A curated collection of papers, benchmarks, surveys and open-source implementations on neural network and model quantization, spanning binary and low-bit networks, post-training and quantization-aware training, large language models, and hardware-efficient deployment.

• Curated lists
• Papers

A native macOS app and command-line tool that runs local language models such as Qwen 3.8 on Apple Silicon using the models' own multi-token prediction heads for faster decoding, exposing an OpenAI- and Anthropic-compatible local server.

• Inference

A local-first personal AI agent that runs on a user's own devices, keeps a long-term memory of their life, browses the web, runs code, and talks over channels such as WhatsApp, then consolidates what it learned into reusable skills while the user is away.

• Agent frameworks
• Agent memory

A JAX-based library for post-training large language models, providing efficient supervised fine-tuning, reinforcement learning and agentic reinforcement learning on TPUs.

• Models
• Inference

An MCP server that lets AI assistants such as Claude control KiCAD for printed circuit board design: authoring and editing schematics, creating custom footprints and symbols, running an autorouter, and pulling in parts data from JLCPCB and LCSC.

• MCP servers

A collection of 273 agent skills and 76 commands across 33 plugins that give Claude Code and Gemini CLI structured workflows for UX research, design systems, UI, interaction and delivery.

• Agent skills
• Design tools

A self-hosted gateway and control plane for MCP servers that connects local and remote servers once, exposes their tools through stable routed endpoints, and adds authentication, per-user credentials, logging and health monitoring on top.

• MCP servers

A Python MCP server, CLI and Claude Code skill for searching and downloading academic papers from sources such as arXiv, PubMed and bioRxiv, favoring free and open sources and returning results standardized for downstream use by a language model.

• MCP servers
• Agent skills

A full-stack TypeScript framework for building, testing and deploying production MCP servers and AI-native apps, with decorator-based tool definitions, dependency injection, built-in authentication and interactive UI widgets attached to tool output.

• MCP servers
• Backend

An MCP server that lets Claude Desktop and other MCP clients control FreeCAD directly: creating and editing 3D models, running Python scripts inside FreeCAD, inspecting open documents, and running finite element analyses.

• MCP servers

An open, terminal-based coding agent, built by a nonprofit community collective rather than a company, that runs on a locally hosted model through Ollama or any OpenAI-compatible API, keeping code and data on the user's own machine.

• Coding agents

A documentation compiler that turns one Markdown source into a website, full-text search, AI-ready context and other knowledge formats together, aimed at serving both human readers and AI agents from the same content.

• Documentation tools

An AI-driven documentation generator, written in Rust, that analyzes a codebase and automatically produces C4-model architecture documentation, including context, container and component diagrams, kept in sync with the code.

• Documentation tools

A collection of TypeScript libraries for working with the Hugging Face API and Hub: managing repositories, calling more than 100,000 hosted inference models, an MCP client, parsers for the GGUF, DDUF and LeRobot formats, and a small library for building tool-using AI agents.

• Inference
• Libraries

The companion code repository for the book AI Agents: The Definitive Guide, with runnable Jupyter notebooks organized by chapter covering agent architectures, planning and reasoning, memory, evaluation, security and production deployment.

• Tutorials
• Books

A local search engine that automatically indexes a folder of code, documents, logs, and PDFs so a coding agent can retrieve exact matches instead of reading files one by one, with an Elasticsearch-compatible API.

• Agent memory
• Databases

A local-first AI agent that runs its control loop and all state on the user's own machine, driving the browser, editing files, running approved shell commands and remembering context across sessions, using small local models through llama.cpp or a cloud model.

• Agent frameworks

An open-source tool that solves hCaptcha challenges in an automated browser by pairing computer-vision models with a multimodal language model, without third-party anti-captcha services or Tampermonkey scripts.

• Browser automation

A support and mirror site for the gyan.dev FFmpeg Windows builds, offering release downloads plus machine-readable files for version numbers and SHA-256 hashes.

• CLI tools