Jia-ren Chang
資深工程師 at 聯發科
Miaoli County, Taiwan, Taiwan
Summary
🤩
Rockstar🎓
Top SchoolJia-ren Chang is a senior engineer with 11 years of experience specializing in computer vision, deep learning, and cognitive neuroscience, currently driving R&D at MediaTek in Taiwan. He holds a PhD in Computer Science and a strong research track record in 3D vision, including publications in well-known conferences and journals. His work on state-of-the-art stereo matching has clear industrial applicability and is reflected in hands-on contributions to the widely used PSMNet repository, where he improved model architecture, disparity regression, and dataset-specific finetuning. Previously as a data scientist at aetherAI, he bridged research and product by adapting advanced vision models for real-world inference scenarios. Jia-ren combines academic depth with practical engineering, frequently moving models from paper to reliable checkpoints and optimized pipelines. His background in neuroscience and psychology gives him a unique perspective on perception-driven ML problems.
11 years of coding experience
2 years of employment as a software developer
Master's degree, Neuroscience, Master's degree, Neuroscience at 國立陽明大學Bachelor's degree, Psychology, Bachelor's degree, Psychology at 中山醫學大學Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at 國立交通大學
Github Skills (13)
mask-rcnn10
computer-vision10
pytorch10
faster-rcnn10
trainings10
modeling10
model-optimization10
preprocess9
python9
preprocessing9
dataprep9
data-preprocessing9
data-loading9
Programming languages (1)
Github contributions (5)
JiaRenChang/PSMNet
Mar 2018 - Sep 2021
Pyramid Stereo Matching Network (CVPR2018)
Role in this project:
ML Engineer
ML EngineerContributions:89 commits, 5 PRs, 86 pushes in 3 years 6 months
Contributions summary:Jia-ren primarily focused on modifying and finetuning the PSMNet model, as indicated by the changes to `finetune.py` and `main.py`. These modifications involve adjusting learning rates, optimizing the 3-px error calculation, and saving model checkpoints. Furthermore, the user made several updates to the core model architecture, including `stackhourglass.py`, and corrected disparity regression in `submodule.py`. The user also adapted the code for different datasets and inference scenarios.
pytorchmatching-networkpyramidstereomatching
JiaRenChang/Batch_Normalized_Maxout_NIN
Apr 2016 - Nov 2017
Contributions:31 commits, 30 pushes, 1 branch in 1 year 7 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.