How to Write to SSDs (VLDB 2026), out-of-place write LeanStore

DeepSeek Harness: Everything is a Plugin.

JSONBench: a Benchmark For Data Analytics On JSON

Terminal session manager for AI coding agents. One TUI for Claude, Gemini, OpenCode, Codex, and more.

White-box LLM post-training course: SFT, DPO, GRPO/RLVR, Agent SFT, and Mini Agent RL on one 16 GB GPU.

A lightweight deep learning training framework implemented from scratch in C++, featuring a PyTorch-style API.

camelAI — an AI coding assistant platform built on Cloudflare Workers and Durable Objects

Large-Scale Disk-Based Vector Index

IMLane Artifact in the Classical Programming era. (Waiting for AI coding enhancement in the future).

Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

A distributed data-processing engine for large-model (LLM) data — SQL, curation, vector/FTS retrieval, multimodal — on stock DuckDB with Rust orchestration.

A cute running cat animation on your macOS menubar.

系统能力大赛编译器作品，获全国一等奖（全国第二名）

旅行行程规划技能：规划 → 小红书调研 → 交互式地图页面 | Agent Skill for trip planning with 小红书 research and interactive map generation

Fast embedded LLM inference

A high-performance inference engine for LLM, VLM, DiT and REC models, optimized for diverse AI accelerators. It is hosted in OpenAtom Foundation.

A tutorial on modern GPU programming for machine learning systems

GenDB, an LLM-Powered Generative Query Engine Built for the Future

2024: 独立开发的第一年，我的成长与收获

Incremental engine for long horizon agents 🌟 Star if you like it!

《金庸群侠传》c++复刻版，已完工

TranGuard 是由华东师范大学数据科学与工程学院数据库系统团队与腾讯云数据库团队联合科研打造的一个面向关系型数据库的事务并发测试工具，通过 LLM 驱动的变异模糊测试（Mutation Fuzzing）自动生成并执行多事务并发场景，检测数据库在不同隔离级别下的正确性。

📰 Must-read papers and blogs on Speculative Decoding ⚡️

LMCache: Supercharge Your LLM with the Fastest KV Cache Layer

Lightweight coding agent that runs in your terminal

Machine Learning Engineering Open Book