Charles Shang
Principal Software Engineer at Roblox
California, United States
Summary
🤩
Rockstar🎓
Top SchoolCharles Shang is a Principal Software Engineer based in California with 11 years of experience building production ML systems and large-scale data pipelines. At Roblox he led projects from voice safety systems and face/voice tracking to pioneering 3D mesh generation and a text- and box-conditioned Cube 1.0 model, and recently built a world-model training platform from scratch. His background blends academic rigor (PhD from Tsinghua) with hands-on engineering—contributing notable open-source work on deformable convolutional networks and a Mask R-CNN implementation that emphasize numerical stability and production readiness. Known for turning research ideas into robust, efficient pipelines, he pairs deep learning research experience with practical system design for real-time, large-scale applications.
11 years of coding experience
5 years of employment as a software developer
Bachelor's degree Computer Science, Bachelor's degree Computer Science at Northwestern Polytechnical UniversityPhD Computer Sciencce, PhD Computer Sciencce at Tsinghua University
Github Skills (16)
mask-rcnn10
object-detection10
faster-rcnn10
pytorch10
coco10
mscoco10
instance-segmentation10
tensorflow10
python10
cython9
cuda9
cprogramming-language7
api-design7
apim7
api7
Programming languages (6)
Github contributions (5)
CharlesShang/DCNv2
Dec 2018 - Apr 2020
Deformable Convolutional Networks v2 with Pytorch
Role in this project:
ML Engineer
ML EngineerContributions:35 commits, 2 PRs, 20 pushes in 1 year 4 months
Contributions summary:Charles primarily focused on improving the functionality and usability of the deformable convolutional networks implementation. Their commits demonstrate debugging efforts, specifically addressing gradient checks and numerical stability issues related to double-precision floating-point calculations. Furthermore, the user added examples and implemented deformable pooling functionality, expanding the capabilities of the library.
pytorchdeformable-convolutional-networkscnn
CharlesShang/FastMaskRCNN
Mar 2017 - May 2017
Mask RCNN in TensorFlow
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:139 commits, 8 PRs, 57 pushes in 1 month
Contributions summary:Charles contributed significantly to the development of a Mask R-CNN implementation using TensorFlow. They added core components, including a COCO dataset converter and an anchor generation module. The user's work involved implementing and integrating several layers and modules, demonstrating an understanding of object detection and instance segmentation. They also refactored and optimized existing code, suggesting a focus on efficiency and performance.
mask-rcnntensorflowmaskrcnnrcnnfasterrcnn
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.