Xun Huang
Founder at Stealth Startup
Pittsburgh, Pennsylvania, United States
Summary
🤩
Rockstar🎓
Top SchoolXun Huang is a founder and PhD-level computer scientist specializing in deep learning and computer vision, with a decade of experience bridging research and production across NVIDIA, Adobe, and academia. He has led generative AI and video world-model efforts as a senior research scientist and taught deep generative models as an adjunct at Carnegie Mellon, while completing doctoral work at Cornell under Serge Belongie. Xun’s open-source contributions include substantive improvements to high-profile NVlabs projects like MUNIT and Imaginaire—ranging from core model architecture and training fixes to MLOps tasks such as Dockerization and domain-invariant perceptual loss. Based in Pittsburgh, he combines hands-on ML engineering with product-minded deployment experience and is now channeling that expertise into a stealth startup. Notably, his background spans both cutting-edge research and practical robustness improvements that directly enhance model performance in real-world pipelines.
10 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree Computer Science, Bachelor’s Degree Computer Science at Beihang UniversityDoctor of Philosophy (Ph.D.) Computer Science, Doctor of Philosophy (Ph.D.) Computer Science at Cornell UniversityExchange Student Computer Science, Exchange Student Computer Science at National University of Singapore
Chinese, English
Github Skills (13)
computer-vision10
pytorch10
image-translation10
docker10
mlops10
deep-learning10
cgan10
python10
dockers10
dcgan10
image-processing10
machine-learning9
tensorflow3
Programming languages (3)
Github contributions (5)
NVlabs/MUNIT
Mar 2018 - Oct 2020
Multimodal Unsupervised Image-to-Image Translation
Role in this project:
ML Engineer
ML EngineerContributions:58 commits, 2 PRs, 16 pushes in 2 years 7 months
Contributions summary:Xun contributed to the `nvlabs/munit` repository, which focuses on multimodal unsupervised image-to-image translation, by updating several Python files including `test.py`, `networks.py`, `trainer.py`, and `utils.py`. These updates suggest involvement in the core model architecture, training procedures, and testing scripts. The code changes include modifications related to network structures, training loops, and data loading, indicating the user's engagement in the model's implementation and refinement.
pytorchdeep-learningtranslationunsupervisedmunit
NVlabs/imaginaire
Sep 2020 - Oct 2020
NVIDIA's Deep Imagination Team's PyTorch Library
Role in this project:MLOps Engineer
MLOps EngineerContributions:7 commits, 1 PR, 7 pushes in 18 days
Contributions summary:Xun primarily contributed to the repository by modifying scripts related to building Docker images for the project, indicating involvement in the deployment and environment setup. They addressed a bug within the MUNIT trainer module and corrected a typo, suggesting a focus on refining model training and ensuring code quality. Furthermore, the user implemented domain-invariant perceptual loss, which directly enhances the model's robustness and performance.
pytorchnvidiaimage-synthesisimage-manipulationdeep-learning
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.