Eugene Vinitsky
Research Scientist at Percepta
Oakland, California, United States
Summary
🤩
Rockstar🎓
Top SchoolEugene Vinitsky is a research scientist and assistant professor with 11 years of experience applying multi-agent reinforcement learning and autonomy research to transportation and large-scale systems. He blends academic rigor—from Caltech and UCSB physics roots to a Berkeley PhD environment—with industrial R&D at Apple, Tesla, DeepMind, and Meta, focusing on sim-to-real, planning under uncertainty, and MARL. Eugene has contributed to high-impact open-source projects like Ray (improving RLlib’s multi-agent and ARS support) and Flow (enhancing vehicle behavior and intersection modeling), demonstrating deep systems and algorithmic expertise. Currently splitting time between NYU Tandon and Percepta, he scales RL methods to real-world, safety-critical domains including autonomous traffic control. He also has entrepreneurial experience building a voice-activated personal-safety app, showing a taste for productizing research. Colleagues describe him as someone who moves fluid theory into robust, production-aware code.
11 years of coding experience
5 years of employment as a software developer
California Institute of TechnologyMaster’s Degree Physics, Master’s Degree Physics at UC Santa Barbara
English, Russian
Stackoverflow
56reputation
512reached
19answers
0questions
Github Skills (29)
system-dynamics10
simulation-modeling10
python10
rllib10
machine-learning10
multi-agent10
object-oriented-programming10
reinforcement-learning10
dynamics10
ray10
ppp9
algorithm9
algorithms9
game-development9
distribute9
Programming languages (7)
Github contributions (5)
eugenevinitsky/sequential_social_dilemma_games
Nov 2018 - Jan 2022
Repo for reproduction of sequential social dilemmas
Role in this project:
Back-end Developer
Back-end DeveloperContributions:4 reviews, 307 commits, 79 PRs in 3 years 2 months
Contributions summary:Eugene appears to be focused on building the core structure and foundational elements of a project related to sequential social dilemma games. They have started the structural setup by defining constants and building out base classes, including a MapEnv class with basic functionality. These foundational tasks appear to be written in Python and they are building the basis for an environment within the stated domain.
sequentialreproductionmachine-learningsocial-dilemmas
flow-project/flow
Feb 2017 - Apr 2022
Computational framework for reinforcement learning in traffic control
Role in this project:Back-end Developer
Back-end DeveloperContributions:9 reviews, 1982 commits, 215 PRs in 5 years 2 months
Contributions summary:Eugene primarily focused on modifying the underlying car following algorithms and vehicle behavior, adjusting the safe distance calculations. They accounted for edge cases in the ring road model and added new functionality for intersections and also improved general aspects of vehicles in the simulation, such as making code more robust. The contributions revolve around the core functionality of the framework, specifically how the vehicles in the model operate.
autonomousreinforcement-learningvehicle-controldeep-reinforcement-learningbenchmark
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.