John Zhang
Software Engineer Of Tensorflow TPU Team at Google
San Jose, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJohn Zhang is a seasoned software engineer with nine years focused on deep learning infrastructure, currently contributing to the TensorFlow TPU team at Google. He combines hands-on systems work across C++ and Python with production ML toolchain design, having led AI framework and inference teams at Black Sesame and Cornami and architected deep-learning products at ASML. His open-source contributions to flagship projects like TensorFlow and JAX show a specialty in compiler and backend work—improving compiler IR APIs, TensorSpec handling, and GDA/jax2tf integration to make TPU workflows more debuggable and interoperable. Trained in applied physics and physics (Shanghai Jiao Tong University, Peking University), he brings a physics-informed, hardware-aware perspective to low-precision inference, quantization, and performance optimization. Colleagues rely on him for bridging silicon, compilers, and high-level ML APIs to deliver efficient, real-time AI deployments.
9 years of coding experience
15 years of employment as a software developer
Master, Physics, Master, Physics at Peking UniversityBS, applied physics, BS, applied physics at Shanghai Jiao Tong University
Stackoverflow
11reputation
871reached
1answer
0questions
Github Skills (16)
mle10
compiler-design10
c-language10
jax10
cprogramming-language10
tensorflow10
python10
ml10
testing9
mlr9
machine-learning9
deep-learning9
xla9
neural-network8
deep-neural-networks8
Programming languages (5)
Github contributions (5)
jax-ml/jax
Jun 2022 - Jan 2023
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:51 reviews, 12 commits, 6 PRs in 6 months
Contributions summary:John primarily contributed to the `jax-ml/jax` repository, focusing on enhancing the jax2tf module. The contributions involved fixing bugs, addressing typos, and adding new tests for GlobalDeviceArray (GDA) functionality within jax2tf. Furthermore, the user improved the handling of TensorFlow functions within JAX, including modifications to `call_tf` and addressing issues related to graph serialization and the treatment of `StatefulPartitionedCall` operations. The work involved changes to core files, including the tests and underlying logic.
pytorchpythonjitautomatic-differentiationgpu
tensorflow/tensorflow
Jan 2023 - Jan 2023
An Open Source Machine Learning Framework for Everyone
Role in this project:ML Engineer
ML EngineerContributions:3 commits, 4 comments in 11 days
Contributions summary:John's commits primarily involve modifying the TensorFlow codebase to support new features and improve existing functionalities related to compiler infrastructure. They focused on enhancing the `experimental_get_compiler_ir` API, enabling support for TensorSpec inputs, and improving its handling of captured inputs, indicating a focus on improving the usability of the compiler's internal representations. Additionally, the user addressed sanitizer errors and implemented a Python API for retrieving tensor specs from tf.Variables, highlighting their work on both the C++ core and Python interfaces. These changes contribute to improving the compiler's capabilities and providing better tools for debugging and optimization within the TensorFlow framework.
pythondata-sciencedeep-learningmlmachine-learning
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.