Shuhei Watanabe
Research Engineer at SB Intuitions
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolShuhei Watanabe is a research engineer based in Tokyo with seven years of experience specializing in AutoML, Bayesian optimization, and scalable hyperparameter tuning. He has contributed to prominent open-source projects like Optuna and Auto-PyTorch, improving backend reliability, visualizations, and training utilities that enhance maintainability and production readiness. His work spans research and engineering roles at AIST, Preferred Networks, and an academic MS at the University of Freiburg, with peer-reviewed publications and awards for accelerating hyperparameter optimization via parallelism and transfer learning. At AIST he was first author on multiple accepted papers and won a competitive domestic award, underscoring his ability to translate research into impactful results. Currently at SB Intuitions, he combines deep theoretical knowledge with pragmatic code refactoring and tooling improvements—evident in test fixes, clearer error messaging, and a flexible LR scheduler added to Auto-PyTorch. Beyond papers and code, he maintains a personal site and demonstrates a habit of improving developer experience as much as model performance.
7 years of coding experience
4 years of employment as a software developer
Master of Science - MS, Computer Science, 1.1/5.0 (1.0 is the best GPA), Master of Science - MS, Computer Science, 1.1/5.0 (1.0 is the best GPA) at The University of FreiburgUniversity of Tokyo
Japanese, English, German
Github Skills (11)
hyperparameter-optimization10
pytorch10
machine-learning10
deep-learning10
automl10
python10
lightgbm10
tabular9
datatable9
visualization8
visualizations8
Programming languages (5)
Github contributions (5)
automl/Auto-PyTorch
Feb 2021 - Jul 2022
Automatic architecture search and hyperparameter optimization for PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:517 reviews, 69 commits, 81 PRs in 1 year 4 months
Contributions summary:Shuhei primarily contributed to refactoring and improving the `auto-pytorch` codebase. Their work focused on modifying error messages, fixing test errors caused by flake8, and refactoring base dataset splitting functions. The user also added a flexible step-wise LR scheduler with minimal changes. The contributions suggest a focus on improving code quality, maintainability, and adding core functionality related to model training.
pytorchhyperparameterdeep-learningoptimizationtabular-data
optuna/optuna
May 2019 - Apr 2025
A hyperparameter optimization framework
Role in this project:Back-end Developer
Back-end DeveloperContributions:639 reviews, 222 PRs, 82 pushes in 5 years 11 months
Contributions summary:Shuhei primarily contributed to the back-end functionality and internal workings of the Optuna project. The commits show a focus on bug fixing, particularly addressing errors related to hyperparameter tuning and the LightGBM integration. The user also introduced code changes related to the visualization modules and added features for customized handling of the code base.
pythonoptimization-frameworkparallelhyperparameteroptimization
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.