Personal AI, On Personal Devices

Open Application-Specific Instruction Set processor tools (OpenASIP)

Mixture-of-experts (MoE) training megakernel for NVL72s

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

CC0 icons for graphics, machine learning, computer vision

ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works…

Tile-Based Runtime for Ultra-Low-Latency LLM Inference

Pythonic binding to the Apple Neural Engine

Use Claude Code, Codex, Pi as subagents or agent in complex workflow.

Local-first session search, analytics, insights, and token use statistics for coding agents, supporting Claude Code, Codex, and more than 20 other agents.

Agentic Kernel Optimization for All — automated GPU kernel optimization for any kernel, any hardware, any language

Rigourous evaluation of LLM-synthesized code - NeurIPS 2023 & COLM 2024

The agent that grows with you

Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1

AI agents running research on single-GPU nanochat training automatically

[MLSys 2026] AccelOpt: Self-improving Agents for AI Accelerator Kernel Optimization

RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI

[KernelGYM & Dr. Kernel] A distributed GPU environment and a collection of RL training methods to support RL for Kernel Generations [ICML 2026]

Running VLA at 30Hz frame rate and 480Hz trajectory frequency

An open-source AI agent that brings the power of Gemini directly into your terminal.

From Minimal GEMM to Everything

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

The open-source alternative to Claude Cowork (powered by opencode)