Seiya Tokui
会社員 at Preferred Networks, Inc.
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolSeiya Tokui is a seasoned software engineer based in Tokyo with 15 years of experience building and stabilizing machine learning and GPU-accelerated systems. He has been at Preferred Networks since 2014 after earlier work at Preferred Infrastructure, blending back-end engineering with DevOps expertise to keep complex ML projects production-ready. His open-source contributions include bug fixes and core improvements to Jubatus’s distributed online learning storage/indexing, CI/CD hardening for ChainerCV, and extending core neural-net functionality in CuPy—work that demonstrates both low-level numerical care and infrastructure reliability. Trained with a master’s in information science from the University of Tokyo, he combines academic rigor with practical engineering, often focusing on the subtle correctness issues (random number generation, indexing, serialization) that make ML systems trustworthy in production.
15 years of coding experience
修士, 情報理工学, 修士, 情報理工学 at 東京大学
Github Skills (27)
docker10
c-language10
python10
scripting10
machine-learning10
bash10
indexing10
dockers10
data-structure10
cicd10
numpy10
script10
computer-engineering10
bit-vector10
data-structures10
Programming languages (5)
Github contributions (5)
jubatus/jubatus
Apr 2012 - Jan 2014
Framework and Library for Distributed Online Machine Learning
Role in this project:
Back-end Developer
Back-end DeveloperContributions:294 commits in 1 year 9 months
Contributions summary:Seiya primarily focused on fixing bugs and improving the codebase related to the storage and indexing components of the Jubatus framework. Their contributions include fixing issues in the `bit_vector` class, correctly implementing the `get_row` function, and resolving a bug related to random number generation within the recommender system. They also addressed issue #37 by implementing a random number generator. These changes indicate a focus on ensuring the correct behavior and stability of the underlying data storage and indexing functionalities within the distributed machine learning framework.
pythoncaffe2data-sciencedeep-learningmachine-learning
cupy/cupy
Dec 2016 - Oct 2019
NumPy & SciPy for GPU
Role in this project:Back-end Developer
Back-end DeveloperContributions:16 releases, 76 commits, 64 PRs in 2 years 10 months
Contributions summary:Seiya's commits primarily focus on refactoring and extending the core functionality of a neural network framework. Their contributions include the addition of features such as a weight decay regularization, a Gaussian and a Bernoulli negative log-likelihood functions. They also added support for serialization, in addition to introducing functions such as Exp and Log. The focus of the commits is on core components related to loss functions, activation functions, and other low-level functionalities of the neural net.
cudapythoncusolvergpunumpy
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.