Wentao Zhu
Research Scientist at wentaozhu91gmail.com
US, China
Summary
👤
SeniorWentao Zhu is a Staff Machine Learning Engineer in Redmond with 11 years of experience bridging cutting-edge research and production ML systems across Amazon, NVIDIA, Kuaishou, and Coupang. He holds a PhD in Computer Science and a strong publication and patent record in medical imaging, multimodal learning, AutoML, and distributed AI, with work like DeepLung contributing to clinically relevant 3D nodule detection. At Amazon he improved rare-object 3D detection and led multimodal content-understanding efforts that drove significant automation and product launches, winning an Amazon Invent Re‑Invent award in CVPR2023. He combines deep research pedigree (NeurIPS, CVPR, MICCAI papers) with pragmatic engineering—shipping models into production and optimizing losses and architectures for real-world gains. Currently focused on LLMs, AGI and generative AI at Coupang, he’s comfortable taking ideas from first-author research to large-scale deployments that cut human effort and cost.
11 years of coding experience
Github Skills (12)
data-preprocessing10
medical-imaging10
computer-vision10
pytorch10
deep-learning10
python10
numpy9
neural-network9
machine-learning9
convolutional-neural-networks9
scipy8
data-augmentation7
Programming languages (7)
Github contributions (5)
wentaozhu/DeepLung
Jan 2018 - Sep 2019
WACV18 paper "DeepLung: Deep 3D Dual Path Nets for Automated Pulmonary Nodule Detection and Classification"
Role in this project:
ML Engineer
ML EngineerContributions:48 commits, 2 PRs, 42 pushes in 1 year 8 months
Contributions summary:Wentao's primary contribution focuses on the development and refinement of the DeepLung model, as indicated by the initial commit and subsequent modifications to the data loading and preprocessing components. The commits demonstrate expertise in adapting the model for lung nodule detection and classification within the medical imaging domain, utilizing data processing techniques and related libraries. The changes to the evaluation scripts suggest the user was also involved in assessing the performance of the model.
3dclassification
wentaozhu/Hierarchical-ELM-Network
Jan 2015 - Mar 2019
IJCNN 2015 Hierarchical extreme learning machine for unsupervised representation learning
Contributions:5 commits, 4 pushes, 5 comments in 4 years 3 months
representation-learningmnist
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.