YoshitomoĀ Matsubara
Research Scientist at Yahoo
Washington, District of Columbia, United States
Summary
š¤©
Rockstarš
Top SchoolYoshitomo Matsubara is a Research Scientist with 11 years of experience at the intersection of machine learning, NLP, computer vision, and ML for science, currently based in Washington, D.C. He holds a Ph.D. in Computer Science from UC Irvine and combines industrial applied-science roles at Amazon and Yahoo with academic research and community service as a multi-year Technical Chair for CVPR, ICCV, and WACV. His open-source work includes developing torchdistill, a widely used PyTorch framework for reproducible knowledge distillation that implements 25+ methods from top conferences, reflecting a strong commitment to reproducibility and benchmarking. Comfortable bridging research and engineering, he has led both founding research engineering at a startup and production-focused applied science at large tech firms. Notably, he pairs deep technical contributions with volunteer leadership in conference program organization and peer review, signaling both technical depth and influence in the ML research community.
11 years of coding experience
8 years of employment as a software developer
Bachelorās Degree Computer Science, Bachelorās Degree Computer Science at Akashi National College of TechnologyMasterās Degree Applied Informatics, Masterās Degree Applied Informatics at University of HyogoUniversity of California, Irvine
Japanese, English
Github Skills (7)
pytorch10
machine-learning10
distill10
dis10
image-classification10
faster-rcnn8
mask-rcnn8
Programming languages (7)
Github contributions (5)
yoshitomo-matsubara/torchdistill
Dec 2019 - Dec 2022
A coding-free framework built on PyTorch for reproducible deep learning studies. PyTorch Ecosystem. š25 knowledge distillation methods presented at CVPR, ICLR, ECCV, NeurIPS, ICCV, etc are implemented so far. š Trained models, training logs and configurations are available for ensuring the reproducibiliy and benchmark.
Role in this project:
ML Engineer
ML EngineerContributions:27 releases, 10 reviews, 1076 commits in 3 years
Contributions summary:Yoshitomo contributed to the development of a knowledge distillation framework built on PyTorch, implementing various knowledge distillation methods. Their work included adding a prototype of the knowledge distillation framework and resolving bugs in existing implementations. The user's commits demonstrate a focus on implementing image classification and machine learning techniques within the PyTorch ecosystem.
knowledge-distillationsemantic-segmentationeccvtrainingbenchmark
yoshitomo-matsubara/hnd-ghnd-object-detectors
Dec 2018 - Mar 2022
[ICPR 2020] "Neural Compression and Filtering for Edge-assisted Real-time Object Detection in Challenged Networks" and [ACM MobiCom EMDL 2020] "Split Computing for Complex Object Detectors: Challenges and Preliminary Results"
Contributions:1 release, 539 commits, 20 PRs in 3 years 2 months
pytorchsplitknowledge-distillationcompressioncoco
Find and Hire Top DevelopersWeāve 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.