Keisuke Fujii
Research Scientist
Summary
🤩
RockstarKeisuke Fujii is an Associate Professor at Nagoya University and founding lead of CyberAgent’s Sports AI Tech Lab, combining over a decade of research and applied work in multi-agent spatiotemporal data science. He specializes in integrating domain knowledge with machine learning, with notable applications in sports analytics (soccer, basketball, handball) and animal group behavior. His background spans academia and national labs (RIKEN), and he brings hands-on ML engineering experience evidenced by contributions to prominent open-source projects like GPflow and xarray, improving core Gaussian process functionality and rolling-window dataset handling. Keisuke also has practical QA and test-automation chops from work on CuPy, reflecting a strong focus on reliable, production-ready scientific software. He holds a PhD from Kyoto University and is known for bridging rigorous research with industry-focused tooling and reproducible code.
10 years of coding experience
Github Skills (22)
unit-testing10
python10
testing10
pandas10
xray10
machine-learning10
ml10
numpy10
tensorflow10
dask10
cupy10
gpflow10
xa10
test-automation10
variational-inference9
Programming languages (9)
Github contributions (5)
pydata/xarray
Feb 2017 - Aug 2020
N-D labeled arrays and datasets in Python
Role in this project:
Back-end Developer & Test Automation Engineer
Back-end Developer & Test Automation EngineerContributions:5 reviews, 69 commits, 124 PRs in 3 years 5 months
Contributions summary:Keisuke's commits primarily focus on improving the functionality and testing of the xarray library, particularly in the context of rolling window operations and dataset handling. They made significant contributions to the `Dataset.rolling` implementation, including adding support and fixing related bugs, and also addressed dimension order restoration after applying rolling windows. Furthermore, they improved the testing coverage of the library by implementing testing for the newly added features. These changes reflect a focus on enhancing the library's core functionality and ensuring the correctness and reliability of its features.
pythonnetcdfnumpydaskarrays
GPflow/GPflow
Jun 2016 - Nov 2016
Gaussian processes in TensorFlow
Role in this project:ML Engineer
ML EngineerContributions:14 commits, 18 PRs, 11 pushes in 4 months
Contributions summary:Keisuke contributed to the GPflow library, a Gaussian process library built on TensorFlow. Their work involved implementing placeholder functionality for fixed parameters and data within GPR and VGP models, which likely aimed at enabling more flexible data handling and model reuse. They also tested the GPR model implementation, verifying its functionality, while noting that VGP was not yet fully functional at the time of these commits. The user's changes touched upon core model components, indicating involvement in the development of fundamental GP functionality.
information-theorygpflowdeep-learningmachine-learningmarkov-chain-monte-carlo
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.