Tsun-yi Yang
Founder And CEO at MIMI AI
City of London, England, United Kingdom
Summary
👤
Senior🎓
Top SchoolTsun-yi Yang is a founder and applied scientist with 11 years of experience building large-scale computer vision and multimodal AI systems, currently leading work on billion-scale operational CV tasks and previously driving retrieval and recommendation research at Meta. He holds a PhD from National Taiwan University and combines deep academic expertise in vision with practical production experience across startups and hyperscalers, including a notable open-source contribution to the CVPR paper FSA-Net for head pose estimation. As founder and CEO of MIMI AI he’s developing a video-to-video generative data platform that emphasizes physical realism and 20x faster generation for robotic training data, reflecting a rare blend of research, product and systems engineering. Comfortable managing cross-team labeling, benchmarking CNN/ViT/VLM models, and synthetic dataset pipelines, he’s equally adept at leading multi-modal LLM research and architecting large training data ecosystems.
11 years of coding experience
11 years of employment as a software developer
Doctor of Philosophy (Ph.D.), Computer Vision, Computer Science and Information Engineering, Doctor of Philosophy (Ph.D.), Computer Vision, Computer Science and Information Engineering at National Taiwan University
English, Chinese, Japanese
Github Skills (9)
mask-rcnn10
keras10
faster-rcnn10
computer-vision10
machine-learning10
deep-learning10
tensorflow10
python10
regression9
Programming languages (8)
Github contributions (5)
shamangary/FSA-Net
Mar 2019 - Oct 2020
[CVPR19] FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation from a Single Image
Role in this project:
ML Engineer
ML EngineerContributions:123 commits, 4 PRs, 120 pushes in 1 year 7 months
Contributions summary:Tsun-yi primarily focused on the development and modification of the FSANET model, a deep learning architecture for head pose estimation. Their commits involve substantial code changes to the `FSANET_model.py` file, indicating direct work on the model's architecture, including modifications to layers, normalization, and activation functions. Additionally, the user updated testing and demo scripts, demonstrating a focus on model evaluation and demonstration.
head-pose-estimationheadtensorflowanglecvpr
shamangary/Pytorch-Stochastic-Depth-Resnet
Feb 2019 - Apr 2019
Contributions:29 commits, 28 pushes, 1 branch in 2 months
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.