Rousslan Dossa
Chief Researcher at Araya Inc.
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolRousslan Dossa is a Chief Researcher at Araya Inc. in Tokyo with nine years of experience building autonomous agents, foundation models, and brain–robot interfaces. He holds a PhD in Information Science (machine learning) from Kobe University and progressed internally from Senior Researcher to Chief Researcher, reflecting rapid technical leadership in applied AI. His research blends deep reinforcement learning and systems engineering—evidenced by an open-source SAC implementation contributed to the popular cleanrl repository that emphasizes research-friendly, single-file RL baselines. Comfortable moving between theory and practice, he develops continuous- and discrete-action RL solutions with features like auto-entropy tuning and multi-Q architectures. Outside formal publications he shares progress and experiments on a personal project page, signaling an experimental, transparent approach to cutting-edge ML problems.
9 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD, Information Science, Machine Learning, Doctor of Philosophy - PhD, Information Science, Machine Learning at Kobe University
Github Skills (8)
deep-reinforcement-learning10
pytorch10
machine-learning10
sac10
ppp10
actor-critic10
python10
reinforcement-learning10
Programming languages (7)
Github contributions (5)
vwxyzjn/cleanrl
Dec 2019 - Oct 2022
High-quality single file implementation of Deep Reinforcement Learning algorithms with research-friendly features (PPO, DQN, C51, DDPG, TD3, SAC, PPG)
Role in this project:
ML Engineer
ML EngineerContributions:97 reviews, 52 commits, 30 PRs in 2 years 10 months
Contributions summary:Rousslan implemented and experimented with a Soft Actor-Critic (SAC) algorithm within the repository, a deep reinforcement learning algorithm. The contributions include the development of the SAC implementation, specifically addressing continuous action spaces, along with support for two Q-functions and a value function. The user also added various features such as auto-entropy tuning and, later, attempted a discrete version of the algorithm, indicating a focus on expanding the algorithm's applicability.
pythondeep-reinforcement-learninggomokutd3reinforcement
dosssman/dosssman.github.io
Feb 2020 - Aug 2024
:triangular_ruler: Jekyll theme for building a personal site, blog, project documentation, or portfolio.
Contributions:32 PRs, 109 pushes, 19 branches in 4 years 6 months
blog-siterulerblog-projectjekylljekyll-theme
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.