Paaras Bhandari
Software Engineer II at Coinbase
New York City Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolPaaras Bhandari is a Software Engineer II based in the New York City area with eight years of experience designing backend systems and distributed architectures, currently building at Coinbase. He has progressed through roles in Wallet and Platform teams, combining production engineering with onboarding and developer experience work. Academically strong with a CS BS and MS from University of Maryland, he pairs rigorous research and teaching experience with hands-on product delivery. An active open-source contributor, he has improved multi-agent RL environments in the popular PettingZoo library by adapting observation spaces and action masks to better support modern RL pipelines. Comfortable both founding startups and shipping at scale, he brings a pragmatic, systems-first approach and a knack for translating academic concepts into production-grade implementations.
8 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Computer Science, 3.82, Bachelor's degree, Computer Science, 3.82 at University of Maryland
English, Hindi
Github Skills (7)
gymnasium10
api10
multi-agent-reinforcement-learning10
python10
reinforcement-learning10
api-design9
numpy9
Programming languages (4)
Github contributions (5)
Farama-Foundation/PettingZoo
Dec 2020 - Dec 2020
An API standard for multi-agent reinforcement learning environments, with popular reference environments and related utilities
Role in this project:
Back-end Developer
Back-end DeveloperContributions:15 commits, 1 PR, 7 pushes in 2 days
Contributions summary:Paaras primarily contributed to the development of environments within the pettingzoo library, focusing on the integration of new observation spaces and action masks, specifically within the classic environments for games like Tic Tac Toe, Connect Four, Backgammon, Checkers, Hanabi, Go, Texas Hold'em, and Chess. The user implemented these changes by modifying the observation spaces to include dictionary-based formats with 'observation' and 'action_mask' keys. These modifications likely aimed to improve the compatibility of these environments with reinforcement learning algorithms and frameworks.
agentreinforcement-learningreinforcement-learning-agentgymnasiumdeep-reinforcement-learning
paarasbhandari/Portfolio-Website
Aug 2017 - Feb 2025
Contributions:2 PRs, 192 pushes, 4 branches in 7 years 6 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.