Peter Zhokhov
Member Of The Technical Staff at OpenAI
Oakland, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPeter Zhokhov is a Member of the Technical Staff at OpenAI with nine years of engineering experience focused on ML platforms, build systems, and backend reliability. He combines a PhD in Physics with hands-on DevOps and backend work, having hardened CI/CD, Docker, and manylinux packaging across high-profile open-source projects like OpenAI Gym and Roboschool. At Sentient Technologies he helped scale ML infrastructure, and his GitHub contributions show a knack for fixing cross-platform build failures and automating complex deployment steps. He has practical experience modifying simulation environments and multi-agent logic in MineRL, bridging low-level Java changes with Python action-space fixes. Based in Oakland, CA, he brings a researcher's rigor to production engineering, often tackling the brittle, behind-the-scenes problems that keep ML systems reproducible and testable.
9 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy (PhD) Physics, Doctor of Philosophy (PhD) Physics at Texas A&M UniversityLomonosov Moscow State University
English, German, Russian
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (34)
dependency-management10
minecraft10
docker10
python10
testing10
machine-learning10
bash10
cmake10
travis-ci10
dockers10
reinforcement-learning10
java10
cicd10
ubuntu10
javas10
Programming languages (5)
Github contributions (5)
openai/roboschool
Nov 2018 - Oct 2019
DEPRECATED: Open-source software for robot simulation, integrated with OpenAI Gym.
Role in this project:
DevOps Engineer
DevOps EngineerContributions:341 commits, 18 PRs, 264 pushes in 10 months
Contributions summary:Peter primarily contributed to the project's build and deployment processes. They created an `install.sh` script to set up dependencies, including building and installing external libraries such as Bullet, Qt, and Boost. The user also configured a deployment script (`deploy.sh`) and made changes to the build configuration files (`config.sh`), demonstrating skills in automating the build process and potentially the deployment to a production environment. Additionally, the user worked on improving and optimizing the build and packaging of wheels for the project, including the manylinux process.
robot-simulationroboticsintegratedreinforcement-learningrobot
openai/gym
May 2018 - May 2021
A toolkit for developing and comparing reinforcement learning algorithms.
Role in this project:DevOps Engineer
DevOps EngineerContributions:22 reviews, 237 commits, 432 PRs in 3 years
Contributions summary:Peter's commits primarily focus on improving the build and testing infrastructure of the `openai/gym` repository. They addressed Travis CI build failures, updated Dockerfiles for different Ubuntu versions, and added/modified dependencies for testing environments. Their work also involved configuring environment variables for MuJoCo, excluding robotics environments when the MuJoCo key was absent, and ensuring the builds worked correctly on various platforms and Python versions. This resulted in a more robust and reliable continuous integration pipeline.
comparingreinforcement-learning-algorithmsdevelopingdeep-learningreinforcement-learning
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.