Enze Xie
Senior Research Scientist at 英伟达公司
Boston, Massachusetts, United States
Summary
🤩
Rockstar🎓
Top SchoolEnze Xie is a Senior Research Scientist based in Boston with nine years of experience advancing computer vision and deep learning, currently focusing on generative AI and acceleration at NVIDIA Research. He holds a PhD from HKU MMLab and has led image and video generation efforts at Huawei Noah’s Ark Lab, contributing to projects like PixArt-Alpha and LLM reasoning work such as AI4Math. Enze has a strong track record in both foundational research and practical system-building—authoring and engineering in well-known open-source repositories like PolarMask and SegFormer and adding production-minded improvements (e.g., deformable convs, FLOPs tooling, and dataset pipelines). His contributions span self-supervised learning, instance segmentation, and vision transformers, and he helped build NVIDIA’s early SegFormer vision-transformer efforts and M2BEV perception for autonomous driving. Notably, he blends deep academic rigor with hands-on engineering across multi-GPU training and benchmarking, making him adept at turning research prototypes into scalable implementations.
9 years of coding experience
2 years of employment as a software developer
硕士 computer science, 硕士 computer science at Tongji University本科 engineering, 本科 engineering at Nanjing University of Aeronautics and AstronauticsThe University of Hong Kong (HKU)
English
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (20)
pytorch10
system-configuration10
python10
imagenet10
semantic-segmentation10
data-preprocessing10
custom-configuration10
deep-learning10
instance-segmentation10
self-supervised-learning10
resnet10
sd-mmc10
computer-vision10
transformer10
configurations10
Programming languages (4)
Github contributions (5)
NVlabs/SegFormer
Jun 2021 - Aug 2022
Official PyTorch implementation of SegFormer
Role in this project:
ML Engineer
ML EngineerContributions:34 commits, 2 PRs, 19 pushes in 1 year 2 months
Contributions summary:Enze primarily contributed to model configuration and dataset integration within the SegFormer project. They made updates to model configurations, particularly for different SegFormer model sizes (B0-B5) and datasets (Cityscapes, ADE20K), adjusting training settings and model parameters. Additionally, the user modified data preprocessing pipelines, ensuring data alignment for efficient model training and performance. They also made some adjustments to the model's FLOPs calculation tool.
pytorchsegformersemantic-segmentationtransformercityscapes
xieenze/PolarMask
Sep 2019 - Nov 2019
Code for 'PolarMask: Single Shot Instance Segmentation with Polar Representation'
Role in this project:ML Engineer
ML EngineerContributions:33 commits, 31 pushes, 1 branch in 1 month
Contributions summary:Enze contributed to a computer vision project focused on instance segmentation. Their commits involved modifying code related to model configurations, including backbone choices (ResNet and ResNeXt), training settings, and dataset configurations. The user also added components, such as Deformable Convolutional Networks, and performed post-processing optimizations, demonstrating a focus on model architecture and performance enhancement.
pytorchrepresentationdeep-learningpolarinstance-segmentation
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.