Peize Sun
Member Of Technical Staff at xAI
Hong Kong Island, Hong Kong, United States
Summary
🤩
Rockstar🎓
Top SchoolPeize Sun is a computer vision and multimodal research engineer with seven years of experience, currently a Member of Technical Staff at xAI working on video generation for Grok Imagine. He holds a PhD from The University of Hong Kong and has contributed to high-profile research at Meta and ByteDance, including multimodal pretraining for Llama4 and autoregressive image/video generation projects. Peize has a strong open-source footprint in detection and tracking—contributing to SparseR-CNN (CVPR2021) and ByteTrack (ECCV-level work)—with hands-on improvements ranging from inference bug fixes to core tracking algorithm implementations. Comfortable bridging research and production, he has a track record of refining models, aligning codebases to conventions, and shipping vision-language features for autonomous driving and segmentation. A detail-oriented engineer, he often focuses on making complex models reliably runnable in real systems, such as fixing focal-loss and input-format issues that unlock reproducible results.
7 years of coding experience
1 year of employment as a software developer
The University of Hong Kong (HKU)Master of Engineering - MEng, Electrical Engineering, Master of Engineering - MEng, Electrical Engineering at Xi'an Jiaotong University
Github Skills (10)
multi-object-tracking10
object-detection10
computer-vision10
pytorch10
python10
kalman-filter10
rgba9
rgb9
mask-rcnn9
faster-rcnn9
Programming languages (4)
Github contributions (5)
PeizeSun/SparseR-CNN
Nov 2020 - Sep 2021
End-to-End Object Detection with Learnable Proposal, CVPR2021
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 2 reviews, 72 commits in 9 months
Contributions summary:Peize primarily contributed to the object detection model within the SparseR-CNN project. They focused on bug fixes related to the inference process and focal loss, and modified input image formats to RGB. This demonstrates a focus on model refinement and ensuring the code runs as intended. The user also updated the codebase to align with existing conventions.
pytorchend-to-endobject-detectioncomputer-visionfaster-rcnn
ifzhang/ByteTrack
Aug 2021 - Oct 2021
[ECCV 2022] ByteTrack: Multi-Object Tracking by Associating Every Detection Box
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:110 commits in 1 month
Contributions summary:Peize implemented `cstrack.py`, `cstrack_new.py` and modified other associated files, indicating a focus on the core tracking functionality. The files suggest implementation of object tracking algorithms, potentially involving Kalman filters and related matching strategies. The changes directly relate to the core purpose of the project (multi-object tracking), suggesting development and enhancement of the tracking algorithms.
pytorchboxdeploymenteccvobject-detection
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.