Shuhei Fujiwara
Software Engineer at DeNA
Chiyoda, Japan
Summary
👤
Senior🎓
Top SchoolShuhei Fujiwara is a software engineer with 11 years of experience in machine learning, TensorFlow, and Google Cloud Platform, currently working at DeNA after earlier roles at Mercari and TopGate. With a master's specialization in mathematical optimization, he focuses on distributed learning and algorithmic optimization for scalable ML systems. As a Google Developers Expert in Machine Learning, he contributes to prominent open-source projects—most notably extending Optuna with TensorFlow pruning hooks and tf.keras callbacks to enable efficient hyperparameter search. He also improves developer tooling and localization, having streamlined Japanese proofreading workflows for TensorFlow docs using RedPen in GitHub Actions. Based in Chiyoda, Japan, Shuhei blends research-grade optimization skills with hands-on engineering to accelerate model training and iteration.
11 years of coding experience
5 years of employment as a software developer
Master's degree, Mathematical Optimization, Master's degree, Mathematical Optimization at Keio University
Stackoverflow
193reputation
11kreached
1answer
10questions
Github Skills (24)
python10
translation10
scripting10
machine-learning10
hyperparameter-optimization10
script10
localization10
tensorflow10
sh10
shell10
testing9
github-ci9
integration-testing9
githubaction-workflow9
documentation8
Programming languages (13)
Github contributions (5)
optuna/optuna
Dec 2018 - Feb 2021
A hyperparameter optimization framework
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 44 commits, 6 PRs in 2 years 2 months
Contributions summary:Shuhei primarily contributed to integrating TensorFlow for pruning in the Optuna hyperparameter optimization framework. Their work involved creating a `TensorFlowPruningHook` for TensorFlow's Estimator, adding corresponding tests, and integrating it with the existing Optuna codebase. Further, the user contributed to the implementation of a `TFKerasPruningCallback` to prune unpromising trials in tf.keras. These additions extend the pruning capabilities of Optuna to TensorFlow and tf.keras, allowing for early stopping and optimization of machine-learning models.
hyperparameter-optimizationpythonmachine-learningparalleldistributed
tensorflow/docs-l10n
Apr 2019 - Aug 2021
Translations of TensorFlow documentation
Role in this project:Localization / Internationalization Specialist
Localization / Internationalization SpecialistContributions:22 commits, 1 PR, 12 comments in 2 years 4 months
Contributions summary:Shuhei primarily focused on improving the Japanese translation workflow for the TensorFlow documentation. Their work involved modifying a shell script (`proofreading.sh`) to incorporate RedPen, a Japanese grammar checker, and streamline the proofreading process within the GitHub Actions environment. They fixed formatting issues, refined the output, and implemented logging for the RedPen tool to improve the efficiency and accuracy of the translations. Additionally, one commit involved translating a data.ipynb notebook.
tensorflowtensorflow-tutorialstranslationlocalization
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.