Fu-hsiang Chan
Algorithm Technical Manager at Ambarella Inc
Hsinchu City, Taiwan, Taiwan
Summary
🤩
Rockstar🎓
Top SchoolFu-hsiang Chan is an Algorithm Technical Manager with a decade of experience designing and shipping computer vision systems for ADAS and autonomous vehicles at Ambarella in Hsinchu. He combines hands-on expertise in model training and optimization—demonstrated by contributions to a TensorFlow Faster R-CNN implementation—with team leadership driving production-ready perception stacks. Academically strong with an MS from National Tsing Hua University (4.15/4.3), he focuses on efficient, real-time vision algorithms tailored for embedded automotive platforms. Known for tuning loss functions and ROI pooling to improve detection performance, he bridges research-grade models and constrained, low-power deployment environments.
10 years of coding experience
5 years of employment as a software developer
Bachelor's degree, Electrical and Electronics Engineering, 3.88 / 4, Bachelor's degree, Electrical and Electronics Engineering, 3.88 / 4 at Yuan-Ze UniversityMaster of Science - MS, Electrical and Electronics Engineering, 4.15 / 4.3, Master of Science - MS, Electrical and Electronics Engineering, 4.15 / 4.3 at National Tsing Hua University
Chinese, English
Github Skills (8)
faster-rcnn10
computer-vision10
machine-learning10
tensorflow10
python10
deep-learning9
algorithm9
algorithms9
Programming languages (1)
Github contributions (5)
smallcorgi/Faster-RCNN_TF
Sep 2016 - Oct 2021
Faster-RCNN in Tensorflow
Role in this project:
ML Engineer
ML EngineerContributions:28 commits, 7 PRs, 31 pushes in 5 years 2 months
Contributions summary:Fu-hsiang primarily contributed to the Faster-RCNN implementation, focusing on modifications within the `lib/fast_rcnn` directory, specifically `train.py`. They also updated the `demo.py` file, demonstrating engagement with the model's usage and demonstration. Their contributions included changes to loss functions and model parameters, indicative of efforts to refine the training process and model performance. Additional commits involved files related to the `roi_pooling_layer`.
deep-learningobject-detectionrcnnfastertensorflow
smallcorgi/3D-Deepbox
Sep 2017 - Sep 2018
3D Bounding Box Estimation Using Deep Learning and Geometry (MultiBin)
Contributions:11 commits, 16 pushes, 1 branch in 1 year
boxdeep-learningboundingestimationgeometry
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.