Tinghui Zhou
Engineering Manager at Roblox
Emeryville, California, United States
Summary
🤩
Rockstar🎓
Top SchoolTinghui Zhou is an engineering manager in Emeryville with 13 years of experience building production-grade generative AI systems, currently leading teams at Roblox focused on 3D avatar generation and hiring ML engineers across levels. He co-founded Humen.ai and led the cross-functional team behind Sway, an AI-driven video synthesis app that reached #3 in the App Store and powered high-profile campaigns including a Super Bowl spot and music-video collaborations. Tinghui pairs deep technical roots from a UC Berkeley PhD and research internships at Google and Disney with hands-on ML engineering—his open-source contributions include enhancements to the influential pix2pix image-to-image translation repo and practical extensions to SfMLearner for monocular depth inference. Comfortable moving between research and product, he has a track record of shipping demos, production pipelines, and data-preparation tooling that bridge prototypes to scalable user experiences. Colleagues describe him as a founder-minded manager who blends technical breadth in computer vision and generative models with product-savvy execution.
13 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of California, BerkeleyBachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at University of MinnesotaMaster of Science - MS, Computer Science, Master of Science - MS, Computer Science at Carnegie Mellon University
Github Skills (15)
computer-vision10
jupyter-notebook10
machine-learning10
deep-learning10
image-generation10
tensorflow10
python10
generative-adversarial-network10
image-processing10
data-pipelines9
data-pipeline9
pytorch9
depth-estimation9
unsupervised-learning8
computer-graphics8
Programming languages (5)
Github contributions (5)
tinghuiz/SfMLearner
Apr 2017 - Oct 2021
An unsupervised learning framework for depth and ego-motion estimation from monocular videos
Role in this project:
ML Engineer
ML EngineerContributions:29 commits, 2 PRs, 21 pushes in 4 years 6 months
Contributions summary:Tinghui's contributions primarily involve modifying and extending the provided code to implement a single-view depth demo, likely for the sfmlearner framework. The user introduces a demo notebook (`demo.ipynb`) that loads and preprocesses an image, then utilizes the `SfMLearner` class to perform depth prediction. The changes encompass setting up the inference graph, restoring a pre-trained model, and running the depth prediction on a sample image. Additionally, the user added code for preparing training data, indicating an involvement in data pipeline development.
pytorchunsupervised-learningdeep-learningunsupervisedvideos
phillipi/pix2pix
Nov 2016 - Apr 2019
Image-to-image translation with conditional adversarial nets
Role in this project:ML Engineer
ML EngineerContributions:12 commits, 1 PR, 9 pushes in 2 years 5 months
Contributions summary:Tinghui primarily contributed to the project by adding and modifying code related to image-to-image translation, which aligns with the project's focus on conditional adversarial networks. Their work includes implementing preprocessing steps for inpainting tasks, adding support for 128x128 images, and integrating Cityscapes evaluation scripts. The user also addressed a minor bug, ensuring a more robust evaluation process.
adversarialtranslationimage-generationcyclegancomputer-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.