Chi Zhang
Research Scientist at ByteDance
Shanghai, Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolChi Zhang is a research scientist with a decade of experience building machine learning systems and production-grade ML infrastructure, currently focused on RLHF at ByteDance in Shanghai. He holds a PhD in Computer Science from USC and a strong academic record across multiple international exchange programs, blending deep theory with pragmatic engineering. Chi contributes to prominent open-source RL tooling—most notably volcengine/verl—where he has optimized RL training pipelines, implemented PPO-related improvements, and scaled workflows in Ray-distributed environments. His background spans research internships (eBay) and high-performing academic results, reflecting an ability to translate complex ML research into robust, efficient training systems. Colleagues describe him as someone who quietly improves validation, metrics, and performance bottlenecks that often go unnoticed until they stop working.
10 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Southern CaliforniaExchange Program in Fall 2014, Electrical Engineering and Computer Science, Exchange Program in Fall 2014, Electrical Engineering and Computer Science at University of California, BerkeleyExchange program supported by China Scholarship Council, School of Information Communication and Technology, Exchange program supported by China Scholarship Council, School of Information Communication and Technology at KTH Royal Institute of TechnologyBachelor of Engineering (B.Eng.), School of Information Science and Engineering, 89.0/100, Bachelor of Engineering (B.Eng.), School of Information Science and Engineering, 89.0/100 at Southeast University
Github Skills (8)
pytorch10
ppp10
distributed-training10
python10
reinforcement-learning10
ray9
machine-learning9
mlops7
Programming languages (5)
Github contributions (5)
volcengine/verl
Oct 2024 - Apr 2025
verl: Volcano Engine Reinforcement Learning for LLMs
Role in this project:
ML Engineer
ML EngineerContributions:322 reviews, 198 PRs, 224 pushes in 5 months
Contributions summary:Chi primarily contributes to the reinforcement learning framework, focusing on improvements to the training pipeline. Contributions include fixing issues related to validation, optimizing performance with gradient checkpointing, and refactoring training-related metrics. The user demonstrates knowledge of reinforcement learning concepts like PPO and its implementation within a Ray distributed environment.
vermouth1992/bracp
Oct 2020 - Oct 2021
Improved Behavior Regularized Offline Reinforcement Learning
Contributions:16 commits, 1 PR, 4 pushes in 1 year
offline-reinforcement-learningimprovedreinforcement-learningdeep-reinforcement-learningreinforcement
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.