Qiao Zhang
Senior Software Engineer at 美国哥伦比亚大学
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolQiao Zhang is a Senior Software Engineer at Google with 12 years of experience combining computer vision research and production-grade ML systems engineering. Based in New York, she has driven core contributions to TensorFlow, JAX, and the XLA compiler—adding FP8 dtypes, optimizing PyTree performance, and refactoring PJRT clients to improve GPU/TPU runtimes. Her background spans academia and industry, including MS work at Columbia and research in pattern recognition, and she teaches data systems and deep learning courses there. At Google she has worked on Daydream/Google Lens teams, translating vision algorithms into scalable infrastructure. Known for reducing GC pressure in deeply nested data structures and improving numerical kernels like expm and segment_sum, she blends algorithmic rigor with low-level runtime optimization. Her combination of open-source impact and production experience makes her adept at bridging research prototypes and performant deployment.
12 years of coding experience
6 years of employment as a software developer
M.S. Electrical Computer Engineering, M.S. Electrical Computer Engineering at Columbia UniversityB. S. Electrical and Computer Engineering, B. S. Electrical and Computer Engineering at Shanghai Jiao Tong University
Chinese, English
Github Skills (36)
c-language10
parallelization10
python10
gpu-programming10
machine-learning10
build-system10
pipelining10
mlr10
ml10
distributed-computing10
numpy10
mle10
compiler-design10
deep-learning10
tensorflow10
Programming languages (5)
Github contributions (5)
jax-ml/jax
Sep 2020 - Jan 2023
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:
Back-end Developer
Back-end DeveloperContributions:5 releases, 180 reviews, 73 commits in 2 years 4 months
Contributions summary:Qiao contributed to the JAX project by updating and modifying build scripts, adding comments to custom derivative rules, and improving the stability of segment_sum. They updated the jaxlib version and made changes to CUDA installation scripts, indicating involvement in build processes and hardware acceleration. Further contributions include making the `expm` function transposable and improving the codebase to work with jax.
pytorchpythonjitautomatic-differentiationgpu
openxla/xla
Sep 2020 - Jan 2023
A machine learning compiler for GPUs, CPUs, and ML accelerators
Role in this project:Back-end Developer
Back-end DeveloperContributions:88 commits in 2 years 4 months
Contributions summary:Qiao made several contributions to the XLA compiler, primarily refactoring the PJRT (Parallel JAX Runtime) client. They focused on modifying and adding new methods to the PjRtClient and PjRtDevice classes, including implementing methods related to infeed and outfeed operations. The user also refactored various aspects of the PJRT client, such as managing buffers and executables, and introduced mechanisms to ensure compatibility with the JAX ecosystem.
compilercommunity-drivenmachine-learningmodular
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.