Kazutaka Morita is a technology leader with 16 years of experience in research and engineering at NTT, now serving as Director after leading teams as a Manager and Research Engineer. He combines deep systems and distributed-storage expertise—contributing to OpenStack Swift fixes around replication, ring building, and proxy streaming—with applied machine-learning compiler work on TVM and nnvm Keras frontends. Comfortable moving between production-grade back-end engineering and ML compiler internals, he has a track record of resolving subtle bugs and adding test coverage for real-world models like MobileNet. Based in Japan, he brings an uncommon blend of enterprise leadership and hands-on open-source contributions that improve both scalability and model portability.
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
ML Engineer
Contributions:1 review, 57 commits, 65 PRs in 2 years 7 months
Contributions summary:Kazutaka made several contributions related to the Keras frontend, implementing support for new features like ReLU6, reshape, and cropping layers. They addressed a bug in the softmax axis and added tests for MobileNet. Further contributions include the addition of multiple outputs, the resolution of padding issues in pooling layers, and improvements to the conversion of GatherNd indices. These changes focused on improving the integration and capabilities of the Keras frontend within the TVM compiler.
Contributions:6 commits, 7 PRs, 8 comments in 29 days
Contributions summary:Kazutaka contributed to the Keras frontend of the nnvm compiler. Their work focused on supporting new features and fixing bugs related to Keras models. Specifically, they implemented ReLU6 support, fixed the softmax axis, and addressed reshape operations within the frontend. Additionally, the user added a test for MobileNet.
cudametalcomputation-graphtvmdeep-learning
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.