Jiayi Weng
Research Engineer at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJiayi Weng is a Research Engineer with nine years of experience building high-performance RL infrastructure and production ML systems, currently working on ChatGPT training and RLHF infra at OpenAI. He has a strong systems background demonstrated by authoring and optimizing EnvPool (C++/pybind11) to achieve over 1M FPS on DGX-A100 and improving multi-agent and data components in the popular Tianshou RL library. Comfortable across C++, Python, and Java/Android, he blends low-level optimization, test automation, and backend data engineering to deliver scalable, well-tested tooling. A Tsinghua undergrad and CMU computational data science master, he also holds a VizDoom competition win and a track record of shipping open-source projects that accelerate RL research.
10 years of coding experience
3 years of employment as a software developer
Bachelor of Engineering - BE, Computer Science and Technology, 3.77/4, Bachelor of Engineering - BE, Computer Science and Technology, 3.77/4 at Tsinghua UniversityMaster's degree, Master of Computational Data Science (System Track), Master's degree, Master of Computational Data Science (System Track) at Carnegie Mellon University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (15)
javas10
c-language10
cprogramming-language10
python10
android-development10
reinforcement-learning10
java10
testing10
data-structures9
ui-design9
data-structure9
numpy8
machine-learning7
cpp7
pybind116
Programming languages (15)
Github contributions (5)
thu-ml/tianshou
Mar 2020 - Dec 2022
An elegant PyTorch deep reinforcement learning library.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:25 releases, 597 reviews, 183 commits in 2 years 10 months
Contributions summary:Jiayi primarily worked on environment and data-related components within the Tianshou reinforcement learning library. They modified code for environment wrappers, data buffers, and collector functionality. These changes involved updates to the core components of the library, adding testing procedures, and incorporating improvements related to handling multi-agent environments and processing data.
deep-reinforcement-learningpytorchpolicy-gradientdqndouble-dqn
sail-sg/envpool
Nov 2021 - Jan 2023
C++-based high-performance parallel environment execution engine (vectorized env) for general RL environments.
Role in this project:Back-end Developer & Test Automation Engineer
Back-end Developer & Test Automation EngineerContributions:21 releases, 136 reviews, 111 commits in 1 year 2 months
Contributions summary:Jiayi contributed to the testing of the `envpool` project, adding a new test suite for partial step functionality and benchmark tests. They implemented and tested a partial step functionality within the atari env and added tests for the implemented functionality in `envpool/atari/atari_envpool_test.py`. They also made adjustments to the `state_buffer_queue.h` and `async_envpool.h` files as part of the testing process.
rl-environmentreinforcement-learningparallel-processingcpp17reinforcement-learning-environments
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.