Szymon Sidor
Technical Fellow at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSzymon Sidor is a Technical Fellow at OpenAI with 12 years of experience building large-scale reinforcement learning systems and production infrastructure. He led development of Rapid, OpenAI’s large-scale RL framework used in high-profile projects like Dota, and progressed through research and engineering roles at OpenAI since 2016. His contributions to the widely used OpenAI Gym include improved Atari benchmarks, reward bounds, and UX enhancements that helped standardize RL evaluation. With an academic background from Cambridge and MIT and prior research work on Schema Networks, he blends deep research instincts with pragmatic systems engineering. Based in San Francisco, he’s comfortable moving between low-level scalable systems and cutting-edge ML research—often improving the tooling that lets agents train at scale.
12 years of coding experience
8 years of employment as a software developer
III highschool in GdyniaMaster of Science - MS Mechatronics Robotics and Automation Engineering, Master of Science - MS Mechatronics Robotics and Automation Engineering at Massachusetts Institute of TechnologyBachelor of Arts (B.A.) Computer Science, Bachelor of Arts (B.A.) Computer Science at University of Cambridge
Polish, English, Spanish
Stackoverflow
89reputation
2kreached
0answers
1question
Github Skills (8)
openai-gym10
benchmarking10
benchmark10
python10
ata10
reinforcement-learning10
atari260010
java6
Programming languages (8)
Github contributions (5)
openai/gym
Nov 2016 - Feb 2018
A toolkit for developing and comparing reinforcement learning algorithms.
Role in this project:
ML Engineer
ML EngineerContributions:14 commits, 19 pushes in 1 year 3 months
Contributions summary:Szymon primarily contributed to the OpenAI Gym repository by adding and modifying Atari benchmarks. These changes included introducing new benchmark configurations for various Atari games (deterministic and stochastic versions), incorporating reward bounds, and adjusting benchmark settings, such as the number of trials and maximum timesteps. Furthermore, the user improved the usability and display of benchmarks on the website, and provided enhancements to include the ability to submit tags. These modifications show the user's work on building benchmarks for reinforcement learning agents to train on.
comparingreinforcement-learning-algorithmsdevelopingdeep-learningreinforcement-learning
siemanko/deeprl
Feb 2016 - Apr 2016
Contributions:20 commits, 4 pushes, 1 branch in 2 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.