Keisuke Umezawa
Director Of Engineering at Mercari, Inc.
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolKeisuke Umezawa is a Director of Engineering based in Tokyo with 11 years of experience building ML-driven data platforms, recommendation and ad-matching systems, and trust & safety infrastructure. He blends hands-on ML modeling and backend engineering with strategic leadership—overseeing budgets, roadmaps, and cross-functional teams at Mercari while driving AI/LLM and data platform initiatives. His background spans finance-grade quantitative systems to productionized customer support automation and content moderation, showing a rare mix of low-latency engineering and applied ML. An active open-source contributor and former Optuna committer, he has improved hyperparameter tuning integrations and contributed test and QA work to GPU and deep-learning libraries, underscoring a strong commitment to reproducibility and code quality. Trained at the University of Tokyo in Mechanical Informatics, he is especially effective at translating research-grade models into reliable, scalable products that materially improve organizational outcomes.
11 years of coding experience
10 years of employment as a software developer
修士, Mechanical Informatics, 修士, Mechanical Informatics at 東京大学
Japanese, Chinese
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (12)
testing10
numpy10
unit-testing10
hyperparameter-optimization10
machine-learning10
unit-test10
python10
lightgbm10
chainer10
test-automation10
cuda9
documentation8
Programming languages (7)
Github contributions (5)
optuna/optuna
Feb 2020 - Jan 2023
A hyperparameter optimization framework
Role in this project:
Back-end Developer
Back-end DeveloperContributions:305 reviews, 237 commits, 155 PRs in 2 years 11 months
Contributions summary:Keisuke's contributions primarily involved modifying the Optuna library to integrate and support the LightGBM tuner functionality. They addressed docstring issues, fixed import statements, and modified code in core files like `lightgbm.py` and `lightgbm_tuner/__init__.py`. The user also made several changes to improve the documentation and address review feedback, indicating a focus on maintaining and improving the library's integration with LightGBM.
pythonoptimization-frameworkparallelhyperparameteroptimization
chainer/chainer
Mar 2017 - Jun 2019
A flexible framework of neural networks for deep learning
Role in this project:Backend Developer & ML Engineer
Backend Developer & ML EngineerContributions:242 commits, 77 PRs, 31 pushes in 2 years 3 months
Contributions summary:Keisuke primarily focused on code style improvements by following autopep8 guidelines. They also made several changes to various testing files, which included testing functionality related to math and normalization tests as well as examples related to different aspects of the project. Additionally, they introduced changes to incorporate the use of the Log2 function within the project.
cudapythonmxnetcaffe2flexible-framework
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.