Hsuan-kung Yang
Senior Research Scientist And Sub Lead at Woven by Toyota
Tokyo, Japan
Summary
👤
Senior🎓
Top SchoolHsuan-kung Yang is a senior research scientist and sub-lead at Woven by Toyota with a decade of experience applying computer vision and machine learning to real-world systems. Currently pursuing a Ph.D. at National Tsing Hua University, he blends academic rigor with production-focused research, having progressed from research scientist to technical lead and now sub-lead within the same advanced mobility group. His hands-on work includes optimizing semantic segmentation models (notably contributions to an ICNet TensorFlow implementation and ADE20k support) and developing vision and reinforcement learning solutions at ELSA Lab. Based in Tokyo, he also has experience teaching CUDA and edge AI at NVIDIA and building practical OCR and analytics pipelines during industry internships, reflecting a knack for translating research into deployable tooling. Colleagues value his rare combination of deep technical depth, leadership across research-to-product transitions, and a steady track record of improving model performance in high-resolution, real-time contexts.
10 years of coding experience
4 years of employment as a software developer
Ph.D., Department of Computer Science, Ph.D., Department of Computer Science at National Tsing Hua University
Github Skills (9)
semantic-segmentation10
computer-vision10
deep-learning10
tensorflow10
model-optimization10
python9
metric8
image-processing8
evaluation8
Programming languages (4)
Github contributions (5)
hellochick/ICNet-tensorflow
Oct 2017 - Jun 2020
TensorFlow-based implementation of "ICNet for Real-Time Semantic Segmentation on High-Resolution Images".
Role in this project:
ML Engineer
ML EngineerContributions:61 commits, 3 PRs, 70 pushes in 2 years 8 months
Contributions summary:Hsuan-kung primarily contributed to the development and improvement of an ICNet model for semantic segmentation. Their work involved updating the `model.py` and `train.py` files, likely refining the network architecture and training parameters. They also improved the evaluation accuracy and modified the inference process as well as added support for the ADE20k dataset. These changes suggest a focus on model optimization, dataset compatibility, and improving the overall performance of the semantic segmentation system.
segmentationsemantic-segmentationdeep-learningresolutionhigh-resolution
hellochick/PSPNet_tensorflow
Oct 2017 - Jun 2019
Contributions:26 commits, 1 PR, 11 pushes in 1 year 7 months
reproducepspnettensorflow
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.