Jun-yan Zhu
Assistant Professor at Carnegie Mellon University
Pittsburgh, Pennsylvania, United States
Summary
🤩
Rockstar🎓
Top SchoolJun-Yan Zhu is an assistant professor of computer science at Carnegie Mellon University with 11 years of experience bridging academic research and production-grade ML systems. His work focuses on image synthesis and understanding—“creating and understanding pixels”—with influential open-source contributions to widely used projects like CycleGAN, pix2pix, BicycleGAN and contrastive unpaired translation. He has a strong industrial research background at Adobe and MIT CSAIL, and his projects have earned awards and visibility in graphics/vision venues. Jun-Yan combines deep theoretical insight from his PhD work with hands-on engineering, frequently improving training pipelines, model architectures, and dataset tooling. He also curates and maintains community resources (e.g., CatPapers) and modernizes codebases for broad reuse, reflecting a pragmatic commitment to reproducibility. Based in Pittsburgh, he blends high-impact research with practical engineering that accelerates adoption of generative vision techniques.
11 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at University of California, BerkeleyBachelor, Computer Science and Technology, Bachelor, Computer Science and Technology at Tsinghua UniversityDoctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at Carnegie Mellon University
Github Skills (26)
pytorch10
python10
machine-learning10
image-translation10
generative-adversarial-network10
deeplearning-ai10
deep-learning10
image-generation10
computer-graphics10
image-manipulation10
html10
neural-network10
computer-vision10
cyclegan10
architecture9
Programming languages (7)
Github contributions (5)
phillipi/pix2pix
Dec 2016 - Aug 2020
Image-to-image translation with conditional adversarial nets
Role in this project:
ML Engineer
ML EngineerContributions:36 commits, 10 PRs, 53 pushes in 3 years 8 months
Contributions summary:Jun-yan's contributions primarily focused on maintaining and refining the model code, with adjustments to the neural network architecture. This includes removing batch normalization layers, fixing comment typos, and updating models, indicating a focus on optimizing and modifying the core machine learning models. They also updated the dataset and model download links and removed unused comments.
adversarialtranslationimage-generationcyclegancomputer-vision
junyanz/pytorch-CycleGAN-and-pix2pix
May 2018 - Dec 2022
Image-to-Image Translation in PyTorch
Role in this project:Backend Developer
Backend DeveloperContributions:176 commits, 80 PRs, 272 pushes in 4 years 7 months
Contributions summary:Jun-yan contributed to the core functionality of the PyTorch-based image-to-image translation project. Their work included implementing a new flag `init_gain` to handle scaling factors in the initialization process, which involved modifications to base options, the CycleGAN model, and the networks module. They also made changes to the pix2pix model and updated dataset download scripts, suggesting a focus on model configuration and data handling within the deep learning project.
pytorchdcgancycleganimage-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.