About Us
AICodingGym is a challenge platform designed to assess and train software engineers' AI-assisted programming skills. As a team of passionate researchers from UC San Diego, we build and maintain a gym environment where users tackle coding challenges that today's AI tools still cannot solve on their own.
Our mission is to train developers to use AI coding tools more effectively while simultaneously benchmarking the performance of those tools. By curating problems at the edge of AI capability, we help the community understand the value of Human-AI collaboration.
Key Contributors
Team: Qiyu Li, Xuan Liu, Shao-Yu Chu, Charlie Sun, Peter Shamoun, Rayirth Dinesh
Supervisors: Haojian Jin, Hao Zhang
Acknowledgements
AI Coding Gym builds on top of several outstanding open-source benchmarks. We gratefully acknowledge:
- SWE-bench Verified — curated real-world software engineering tasks from GitHub issues
- SWE-bench Pro — professional-grade software engineering challenges
- MLE-bench — machine learning engineering benchmark from Kaggle competitions
- Code Review Benchmark — benchmark for evaluating code review capabilities