AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary

DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.

Self-Hostable MicroVMs for your Agents

Inference at the speed of light.

Open-source inference server and production cluster for all the models your agent needs.

TorchLean is the first unified Lean 4 framework for neural-network specification, execution, and verification.

A unified storage SDK for object and blob backends. One small, honest API. Web-standards I/O.

Computational Graphics - THU Spring 2018

AssetOpsBench - Industry 4.0: A unified benchmark and framework for building, orchestrating, and evaluating domain-specific AI agents for Industry 4.0 asset operations and maintenance, with 460+ sc…

agent multiplexer that lives in your terminal.

The Doubleword Inference Stack is the easiest & most performant way to run genAI infrastructure in your private environment.

✨ Cross-agent skills that help coding agents use Entire context from Checkpoints, sessions, and git history to search past work, explain code, and hand off sessions.

Agent OS: Stop prompting. Start specifying.

Headless CLI client for stateful Agent Client Protocol (ACP) sessions

(Mirror) S3-compatible object store for small self-hosted geo-distributed deployments. Main repo: https://git.deuxfleurs.fr/Deuxfleurs/garage

The open-source app everyone uses to manage agents at work

On-device Speech AI for Apple Silicon

A hybrid programming language combining Lean4's formal verification with blazing-fast compilation, actor-based agent orchestration, AI-driven optimization, and vector-backed agent memory.

amdgpu example code in hip/asm

A skill for thinking

🦄 ai that works - every tuesday 10 AM PST

Machine Learning Engineering Open Book

🤗 smolagents: a barebones library for agents that think in code.

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

Generating Efficient AI-Centric Kernels

Verified tensor graph optimization in Lean 4: constructive soundness proofs + equality saturation + verified extraction via e-graph↔circuit bijection + multi-target code generation.

Verified GPU programming framework for Lean 4. Write type-safe WebGPU shaders with formal verification, hardware-accelerated matrix ops, and cross-platform support (Metal/Vulkan/D3D12). Build prova…

Row-wise block scaling for fp8 quantization matrix multiplication. Solution to GPU mode AMD challenge.

A lightweight multi-GPU inference engine for LLMs on mid/low-end GPUs.