Zhicheng Yan
Principal Research Scientist Director at Meta
Menlo Park, California, United States
Summary
👤
Senior🎓
Top SchoolZhicheng Yan is a Principal Research Scientist and Director at Meta with 11 years of experience building on-device perception systems and leading research in 3D generative AI and spatial intelligence for VR/AR/XR products. He progressed from hands-on contributions at Facebook AI to senior leadership roles in Reality Labs, managing teams focused on object and scene understanding in egocentric data. Technically rooted in computer vision, machine learning and computer graphics (PhD, UIUC), he has driven large-scale video and image understanding systems in production and improved core tooling like PyTorch Vision and Detectron2. Zhicheng combines deep research output with pragmatic engineering—his open-source work includes enhancing video processing robustness and dataset/sampler improvements in widely used repositories. Based in Menlo Park, he pairs academic rigor with product-focused delivery, often optimizing for on-device efficiency and real-time constraints. An underappreciated strength is his knack for translating foundational 3D research into scalable, low-latency components for consumer experiences.
11 years of coding experience
12 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Computer Vision Machine Learning Computer Graphics, Doctor of Philosophy (Ph.D.) Computer Vision Machine Learning Computer Graphics at University of Illinois Urbana-ChampaignMaster’s Degree computer science, Master’s Degree computer science at Zhejiang University
Chinese, English
Github Skills (20)
pytorch10
subsampling10
python10
video-processing10
machine-learning10
sampling10
detectron10
resampling10
computer-vision10
ffmpeg9
visualization9
data-visualizations9
data-visualisation9
visualizations9
segmentation9
Programming languages (3)
Github contributions (5)
facebookresearch/detectron2
Jun 2021 - Mar 2024
Detectron2 is a platform for object detection, segmentation and other visual recognition tasks.
Role in this project:
ML Engineer
ML EngineerContributions:7 PRs in 2 years 9 months
Contributions summary:Zhicheng contributed to the Detectron2 project by implementing and modifying features related to model visualization, data sampling, and performance optimization. They added functionalities like controlling color jittering in visualizations and adjusting font size scaling. Furthermore, they modified data samplers by implementing the option to change the per-category weight calculation method and logging of sampling parameters. They also addressed data type inconsistencies related to resizing transformations.
detectron2deep-learningface-recognitionrecognitionobject-detection
pytorch/vision
Sep 2019 - Oct 2019
Datasets, Transforms and Models specific to Computer Vision
Role in this project:Back-end Developer
Back-end DeveloperContributions:22 commits, 28 PRs, 27 comments in 1 month
Contributions summary:Zhicheng's primary contributions focused on enhancing the video processing capabilities within the PyTorch vision library. They implemented a video reader inception commit, which included adding and modifying methods for video metadata handling, bug fixes, and improving robustness. The user addressed compatibility issues, refactored code related to video transforms, and integrated fast video probing functionality. Furthermore, they made improvements to the video dataset classes, and refined existing video samplers.
pytorchvisiondeep-learningdatasetcomputer-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.