Joe Wu
CEO And CTO at Qubot Technology
Singapore
Summary
👤
Senior🎓
Top SchoolJoe Wu is a tech entrepreneur and research-driven CTO with 12 years of experience building AI-powered medical products now deployed in over 300 hospitals and leading Qubot Technology as CEO and CTO. He has more than 60 invention patents and 40+ academic papers with 700+ citations, and previously worked with leaders in the field including Yoshua Bengio, Microsoft Research and A*STAR on speech recognition, NLP and deep learning. At BioMind he scaled a team from zero to 250 clinicians and engineers to deliver six assistive diagnosis and treatment products, combining deep research rigor with product delivery. A practical performance engineer as well as researcher, he has contributed optimized CUDA kernels and preprocessing pipelines to prominent Torch7 repositories, showing hands-on impact on ML tooling. He also serves as Vice Chairman of AI branches in two national-level medical associations, bridging academic leadership and clinical deployment.
12 years of coding experience
7 years of employment as a software developer
Bachelor of Science Honours Double Majors in Physics and Mathematics, Bachelor of Science Honours Double Majors in Physics and Mathematics at National University of SingaporeMaster's degree Computer Science, Master's degree Computer Science at McGill University
Github Skills (15)
tensorrt10
data-preprocessing10
cuda10
computer-vision10
pytorch10
machine-learning10
tensor10
deep-learning10
tensorflow10
parallel-computing10
performance-optimization10
operation10
cprogramming-language9
cifar1009
c-language9
Programming languages (6)
Github contributions (5)
nicholas-leonard/dp
Nov 2013 - Aug 2014
A deep learning library for streamlining research and development using the Torch7 distribution.
Role in this project:
ML Engineer
ML EngineerContributions:31 commits in 9 months
Contributions summary:Joe implemented the ZCA whitening and Global Contrast Normalization (GCN) preprocessing techniques for the deep learning library. They added the ZCA module and a LeCunLCN and GCN Preprocess and also updated and tested the existing preprocessor for Standardize and GCN. Furthermore, they integrated these preprocessing steps within the CIFAR10 dataset and also added data loading and preprocessing functionalities for the SVHN dataset.
pytorchdeep-learningresearch-and-developmentcomputer-visionmachine-learning
torch/cutorch
Jun 2014 - Jun 2014
A CUDA backend for Torch7
Role in this project:Back-end Developer / Performance Engineer
Back-end Developer / Performance EngineerContributions:7 commits in 7 days
Contributions summary:Joe focused on optimizing the `indexSelect` function within the CUDA backend for Torch7. Their work involved implementing a CUDA kernel for `indexSelect`, adding speed comparisons, and fixing related test issues. The user's contributions include optimizing existing CUDA code for better performance and expanding functionality through the addition of `indexFill` and `indexCopy` functions. These changes indicate an emphasis on improving the efficiency of tensor operations within the CUDA environment.
cudagpubackendcuda-backend
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.