Michael Petrov
Member Of Technical Staff at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolMichael Petrov is a versatile Member of Technical Staff with 14 years of engineering experience spanning iOS, C/C++, Linux, AWS, and full-stack backend work. Currently on the OpenAI Five team in San Francisco, he brings a game-AI and systems perspective informed by prior mobile-platform leadership at Dropbox and co-founding a high‑scale consumer app (Couple) with millions of downloads. He has deep low-level and performance expertise—Core Data tuning, concurrency, real‑time protocols and custom networking—combined with cloud architecture skills (Node.js, AWS) that delivered 99.9% API uptime in production. Michael is also an active ML engineer contributor to the well-known lucid repository for neural network interpretability, where he improved batching, stability, and model import usability—an unusual bridge between mobile/infra craftsmanship and ML research tooling.
14 years of coding experience
12 years of employment as a software developer
Y CombinatorToronto Metropolitan UniversityUniversity of Waterloo
Stackoverflow
2,257reputation
268kreached
59answers
1question
Github Skills (20)
interpretation10
python10
machine-learning10
tensorflow10
jupyter-notebook9
socket9
visualization8
visualizations8
api-design7
apim7
api7
tcp-client6
php6
iphone6
tcp6
Programming languages (8)
Github contributions (5)
tensorflow/lucid
Jul 2019 - Feb 2020
A collection of infrastructure and tools for research in neural network interpretability.
Role in this project:
ML Engineer
ML EngineerContributions:49 commits, 20 PRs, 21 pushes in 7 months
Contributions summary:Michael primarily contributed to the core functionalities of the "lucid" repository, a project focused on neural network interpretability. Their commits demonstrate a focus on improving the codebase related to objectives, graph analysis, and image parametrization, as well as enhancing the usability and flexibility of the model import process. Notably, the user addressed a division-by-zero error and implemented batching support, improving the overall stability and functionality.
pytorchinterpretabilitydeep-learningjupyter-notebookinfrastructure
tenthbitinc/socket.IO-objc
Feb 2012 - May 2014
Contributions:14 commits in 2 years 3 months
socketsocket-iowebsocketnodejsios
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.