Kentaro Wada
Computer Vision Sub-Lead at Mujin
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolKentaro Wada is a computer vision and robotics engineer with 12 years of experience, currently serving as Computer Vision Sub-Lead at Mujin in Tokyo. He combines deep learning research (PhD from Imperial College London) with hands-on systems work—shipping ROS packages, image pipelines, and production-ready tools for perception and robot learning. An active open-source contributor, he maintains popular projects like gdown and labelme and has contributed to core libraries such as Chainer, CuPy and PCL, showing expertise from low-level GPU array ops to high‑level annotation and visualization tooling. Kentaro focuses on applying ML to practical products, improving developer UX and robustness (e.g., caching, dynamic reconfigure, and improved rviz/rqt plugins), and often bridges research prototypes to deployable services. Notably, his background spans both web/data roles and drone-based warehouse inspection, reflecting a talent for turning varied domain knowledge into applied automation.
12 years of coding experience
8 years of employment as a software developer
Science, General Science, Science, General Science at Zeze High School of Shiga PrefectureDoctor of Philosophy - PhD, Computing, Doctor of Philosophy - PhD, Computing at Imperial College LondonUniversity of Tokyo
Japanese, English
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (85)
tooling10
opencv10
python10
package-management10
toolbox10
image-processing10
testing10
data-manipulation10
command-line-interface10
point-cloud-library10
deep-learning10
faster-rcnn10
cpp10
visualizations10
ros10
Programming languages (19)
Github contributions (5)
wkentaro/gdown
Oct 2015 - Dec 2022
Google Drive Public File Downloader when Curl/Wget Fails
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:11 releases, 10 reviews, 332 commits in 7 years 2 months
Contributions summary:Kentaro significantly improved the user interface and overall functionality of the gdown tool. They added features like a quiet mode, support for Python 3, and the ability to specify an output file. Furthermore, they updated the README with examples and installation instructions, also integrating badges and supporting multiple URLs. They refactored the code and improved the caching mechanism.
pythonfailswgetsecuritygoogle-drive
wkentaro/pytorch-fcn
Feb 2017 - Jun 2022
PyTorch Implementation of Fully Convolutional Networks. (Training code to reproduce the original result is available.)
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:316 commits, 45 PRs, 247 pushes in 5 years 4 months
Contributions summary:Kentaro implemented FCN32s, a fully convolutional network, in PyTorch, demonstrating a strong understanding of deep learning architectures. They incorporated a VGG16 model for transfer learning, suggesting experience with model optimization and leveraging pre-trained weights. The user fixed various issues, including validation and loss calculation, indicating experience in debugging and refining the model's functionality.
pytorchefficientnetdeep-learningconvolutionalconvolutional-networks
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.