Kaiwen Wang
Researcher at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolKaiwen Wang is a researcher and PhD candidate in computer science at Cornell Tech with nine years of experience building and deploying reinforcement learning and LLM systems. Currently at OpenAI working on safety RL, Kaiwen has held research internships at Google, Netflix, and Microsoft focused on LLM reasoning, post-training value alignment, and learning from execution feedback. Previously a core RL engineer at Facebook AI, they improved production metrics with offline RL and contributed performance-critical back-end work to the widely used facebookresearch/ReAgent repository by vectorizing replay buffers. Their background blends rigorous academic training with hands-on systems engineering across distributed RL infrastructure and alignment research. Based in San Francisco, Kaiwen combines deep algorithmic understanding with practical optimization skills that bridge research prototypes to production. An understated strength is their repeated success translating complex theoretical ideas into efficient, production-ready implementations.
9 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy - Ph.D., Computer Science, Doctor of Philosophy - Ph.D., Computer Science at Cornell TechBachelor of Science - BS, Mathematics and Computer Science, Bachelor of Science - BS, Mathematics and Computer Science at Carnegie Mellon University
English, French, Chinese
Github Skills (7)
pytorch10
python10
reinforcement-learning10
numpy9
performance-optimization9
machine-learning9
algorithms8
Programming languages (2)
Github contributions (5)
facebookresearch/ReAgent
Mar 2020 - Aug 2021
A platform for Reasoning systems (Reinforcement Learning, Contextual Bandits, etc.)
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:89 commits, 85 PRs, 31 branches in 1 year 5 months
Contributions summary:Kaiwen made several significant contributions to the `facebookresearch/reagent` repository, focusing on improving and optimizing the replay buffer functionality. Their work involved vectorizing the replay buffer, which led to increased speed compared to the original iterative sampling method. This included modifications to the ReplayBuffer class, specifically within the ml/rl/replay_memory/circular_replay_buffer.py file, indicating a focus on reinforcement learning systems and model performance.
reinforcement-learningcontextualbanditscontextual-banditsreinforcement
kaiwenw/kaiwenw.github.io
Jun 2017 - Sep 2022
Contributions:90 pushes, 1 branch in 5 years 3 months
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.