Masaki Kozuki
Software Engineer at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolMasaki Kozuki is a software engineer with 10 years of experience specializing in deep learning frameworks and GPU-accelerated libraries, currently working at NVIDIA on PyTorch foreach functions, fused optimizers and apex.transformer. He is an active open-source contributor to flagship projects like PyTorch, Chainer, CuPy and Optuna, where his work ranges from CUDA performance fixes and low-precision training to refactoring and adding practical ML features such as Hyperband and pruning callbacks. Masaki blends systems-level CUDA/kernel understanding with higher-level ML engineering, frequently improving performance-critical code paths and memory integrations (DLPack, NULL strides) that many users rely on. His background includes recurring part-time and internship roles at Preferred Networks where he helped productionize Chainer/ChainerCV and Optuna, and a string of applied ML internships focused on edge inference and drug-discovery models. Based in California and academically trained at the University of Tokyo (Magna cum laude), he brings both rigorous research instincts and pragmatic maintenance-focused craftsmanship to complex ML codebases. A less obvious strength is his consistent focus on code hygiene—typo fixes, doc improvements and refactors—that materially raises long-term project stability and developer productivity.
10 years of coding experience
5 years of employment as a software developer
学士, 教養, 学士, 教養 at 東京大学
Github Skills (30)
pytorch10
python10
model-driven10
chainer10
machine-learning10
numpy10
hyperparameter-optimization10
model-building10
pruning10
deep-learning10
cupy10
cuda10
computer-vision10
modeling10
model-driven-development10
Programming languages (11)
Github contributions (5)
optuna/optuna
Mar 2019 - Mar 2021
A hyperparameter optimization framework
Role in this project:
Data Scientist & ML Engineer
Data Scientist & ML EngineerContributions:287 reviews, 529 commits, 186 PRs in 2 years
Contributions summary:Masaki primarily contributed to the development of the hyperparameter optimization framework Optuna, focusing on enhancements related to pruning and integration with machine learning models. Their work involved refactoring existing code, addressing documentation issues, and adding examples for integration with various machine learning frameworks such as PyTorch and FastAI. The user demonstrated expertise in implementing and testing pruning callbacks for efficient optimization in complex machine learning workflows.
pythonoptimization-frameworkparallelhyperparameteroptimization
chainer/chainer
Feb 2017 - Oct 2019
A flexible framework of neural networks for deep learning
Role in this project:Back-end Developer
Back-end DeveloperContributions:251 commits, 77 PRs, 342 comments in 2 years 8 months
Contributions summary:Masaki's contributions primarily involve code refactoring and correcting typos, specifically within the Chainer framework's core functionalities. This includes modifications to the `functions/normalization/batch_normalization.py`, `reporter.py` file and the integration of VGG19 model. The commits demonstrate a focus on code maintenance, improving code readability, and updating code to reflect best practices 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.