Yuya Unno
リテールソリューションズ事業本部 本部長 at Preferred Networks, Inc.
Chiyoda, Japan
Summary
🤩
Rockstar🎓
Top SchoolYuya Unno is a seasoned AI and software leader with 15 years of experience, currently heading Retail Solutions at Preferred Networks in Tokyo. He rose through technical and executive roles at Preferred Networks and its predecessor, blending hands-on research with strategic product leadership. A former researcher at IBM, he holds an MS in Computer Science from the University of Tokyo and has deep expertise in NLP and deep learning. As an active open-source contributor, he helped develop core features in the influential Chainer framework and optimized attention and decoder logic in the ESPnet speech toolkit, showing a knack for both algorithmic clarity and performance tuning. He combines research-grade rigor with product execution, often surfacing engineering improvements that simplify implementations while boosting efficiency. Colleagues rely on him to bridge cutting-edge model development and practical deployment in retail and speech applications.
14 years of coding experience
8 years of employment as a software developer
BS, Information Science, BS, Information Science at 東京大学MS, Computer Science, MS, Computer Science at 東京大学 / The University of Tokyo
Github Skills (19)
pytorch10
python10
machine-learning10
numpy10
deeplearning-ai10
deep-learning10
neural-network10
cuda10
speech-recognition10
testing9
machine-translation9
attention-mechanism9
tensorflow8
speech-synthesis8
chainer8
Programming languages (13)
Github contributions (5)
chainer/chainer
Apr 2015 - Feb 2019
A flexible framework of neural networks for deep learning
Role in this project:
ML Engineer
ML EngineerContributions:4 releases, 2629 commits, 1234 PRs in 3 years 10 months
Contributions summary:Yuya contributed extensively to the Chainer deep learning framework. Their work focused on enhancing and optimizing core functionalities, specifically by implementing the log-softmax activation function, introducing tests for the newly implemented functionality, and fixing potential issues and bugs in existing code related to existing activation functions and layer definitions. The user also introduced the ability to handle variable length inputs and a tool to generate a model's architecture.
cudapythonmxnetcaffe2flexible-framework
espnet/espnet
Dec 2017 - Jan 2018
End-to-End Speech Processing Toolkit
Role in this project:ML Engineer
ML EngineerContributions:6 PRs in 17 days
Contributions summary:Yuya primarily contributes to the end-to-end speech processing toolkit by optimizing and refactoring code related to attention mechanisms and decoder implementation. Their changes involve using efficient functions like `cumsum` and `xp.full`, simplifying operations using `reshape` and `flatten`, and updating attention calculations with `broadcast_to` and `separate`. These modifications appear focused on enhancing model performance and code clarity within the context of the speech recognition and synthesis tasks.
speech-recognitionspeech-separationchainerspoken-language-understandingspeech-processing
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.