Takuma Seno
Staff Researcher at Turing Inc
Minato, Japan
Summary
🤩
Rockstar🎓
Top SchoolTakuma Seno is a Staff Researcher with 11 years of experience, holding a Ph.D. in computer science from Keio University and currently driving end-to-end deep reinforcement learning research for level 5 autonomous driving at Turing Inc. He was a core contributor to Sony AI’s Gran Turismo Sophy project and has a strong track record implementing and refining neural network libraries and RL tooling. His open-source work includes backend and DevOps improvements to d3rlpy and core computation-graph features for nnabla, reflecting a blend of research rigor and production-focused engineering. Prior roles span ML research and robotics, full-stack web development, and cloud architecture, giving him rare cross-domain fluency from backend systems to real-world autonomous systems. Colleagues rely on him to translate advanced RL methods into robust, reproducible code and scalable training pipelines.
11 years of coding experience
11 years of employment as a software developer
Ph.D, Computer Science, Ph.D, Computer Science at Keio University
Chinese
Github Skills (14)
neural-network10
machine-learning10
c-language10
cprogramming-language10
python10
reinforcement-learning10
pytest9
github9
documentation9
testing9
apidoc9
api9
git-repository9
git9
Programming languages (8)
Github contributions (5)
takuseno/d3rlpy
Jun 2020 - Jan 2023
An offline deep reinforcement learning library
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:31 releases, 96 reviews, 1268 commits in 2 years 8 months
Contributions summary:Takuma's contributions focused on adding docstrings and fixing an example related to the Fitted Q-Evaluation (FQE) off-policy evaluation method within the d3rlpy library. They also implemented a "save_metrics" flag to disable logging and modified various aspects of the online training functionalities, including the modification of parameters, indicating the user's involvement in improving the library's functionality and documentation. Additionally, the user updated the reproduction scripts.
pytorchdeep-learningreinforcement-learningdeep-reinforcement-learningoffline-rl
sony/nnabla
Sep 2018 - Dec 2019
Neural Network Libraries
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:41 commits, 17 PRs, 36 comments in 1 year 3 months
Contributions summary:Takuma contributed significantly to the `nnabla` repository, a neural network library. Their work involved implementing core functionalities related to computation graphs, specifically enabling the `forward_all` function and its Python interface, which likely improved the library's execution capabilities. Furthermore, they added an `assign` function, enhancing the library's array manipulation capabilities. These changes suggest a focus on extending and refining the library's core features for efficient neural network computations.
deep-learningneural-networksmachine-learningneural-networktensorflow
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.