Ting-chun Wang
Principal Research Scientist at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolTing-chun Wang is a Principal Research Scientist at NVIDIA with nine years of experience at the intersection of machine learning, computer vision, computational photography, and graphics. He leads generative AI work for the NVIDIA Picasso product and has a strong track record of translating research into production-ready deep learning systems. His contributions to high-profile open-source projects like vid2vid, few-shot-vid2vid, and pix2pixHD show hands-on expertise in photorealistic video-to-video translation, optical flow integration (FlowNet2), SPADE and attention-based generators, and dataset tooling. A PhD-trained researcher from UC Berkeley, he blends academic rigor with practical engineering—having previously built interactive neural-rendering tools at Adobe and simulation systems in academia. Based in California, he is skilled at adapting state-of-the-art models for high-resolution, real-time applications and resolving cross-version and code-quality issues in collaborative codebases. Colleagues know him for improving core model architectures while also tending to the less-visible but critical plumbing that makes research reproducible and deployable.
9 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Electrical Engineering and Computer Science, Doctor of Philosophy (Ph.D.) Electrical Engineering and Computer Science at University of California, BerkeleyBachelor's degree Electrical and Electronics Engineering, Bachelor's degree Electrical and Electronics Engineering at National Taiwan University
Chinese, English
Stackoverflow
3reputation
557reached
0answers
1question
Github Skills (17)
pytorch10
python10
machine-learning10
generative-adversarial-network10
deep-learning10
deep-neural-networks10
optical-flow10
neural-network10
computer-vision10
faster-rcnn8
mask-rcnn8
attention-mechanism8
uibutton6
iphone6
uiimageview6
Programming languages (2)
Github contributions (5)
NVIDIA/vid2vid
Aug 2018 - Jul 2019
Pytorch implementation of our method for high-resolution (e.g. 2048x1024) photorealistic video-to-video translation.
Role in this project:
ML Engineer
ML EngineerContributions:38 commits, 17 pushes, 1 branch in 10 months
Contributions summary:Ting-chun's primary contribution involved integrating and configuring FlowNet2, a pre-trained optical flow network, within the project. They added scripts to download and install the FlowNet2 model, including specific support for PyTorch 0.4.1. Further modifications to the code base were made to incorporate FlowNet2 into the model architecture, specifically within the models and scripts directories. In addition, the user addressed several flake8 issues, updating the data, models and temporal datasets.
pytorchtranslationmethodresolutioncomputer-vision
NVIDIA/pix2pixHD
Dec 2017 - Jun 2019
Synthesizing and manipulating 2048x1024 images with conditional GANs
Role in this project:ML Engineer
ML EngineerContributions:26 commits, 7 PRs, 24 pushes in 1 year 6 months
Contributions summary:Ting-chun's contributions center around modifying and enhancing the neural network models used in the pix2pixHD project, as evidenced by the code changes in `models/networks.py`. The user is likely involved in feature encoding and generation processes within the model, as shown by adding explanations and making updates to the dataset, including the UI model. This work involves adapting the model for new datasets and addressing potential version issues.
pytorchdeep-learningconditionalcomputer-graphicscomputer-vision
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.