Ivy Zheng
Senior Software Engineer at Google DeepMind
California, United States
Summary
🤩
Rockstar🎓
Top SchoolIvy Zheng is a Senior Software Engineer based in California with 11 years of experience building highly scalable, distributed systems and infrastructure, with deep expertise in machine learning tooling. Currently at Google DeepMind, she actively maintains core ML libraries JAX and Flax and has contributed notable features like multi-host checkpointing for GlobalDeviceArrays and asynchronous checkpoint saving. Previously she worked on YouTube Search infrastructure and ML-powered ranking, bringing production-grade reliability to large-scale indexing and ranking pipelines. Her background spans research and industry—implementing AI systems at Northwestern and visualization/data tools at Facebook—demonstrating a blend of algorithmic rigor and practical engineering. Beyond production systems, Ivy’s open-source work shows a focus on robustness and serialization in high-performance ML stacks, an often overlooked but critical part of scalable model deployment.
11 years of coding experience
7 years of employment as a software developer
Master of Science - MS, Computer Science, Master of Science - MS, Computer Science at Northwestern University
English, Chinese
Github Skills (14)
tree10
data-tree10
machine-learning10
tree-structure10
data-serialization10
jax10
python10
serialization10
async9
numpy9
asynchronous9
merge-conflicts8
debugging7
debug7
Programming languages (4)
Github contributions (5)
google/flax
May 2022 - Jan 2023
Flax is a neural network library for JAX that is designed for flexibility.
Role in this project:
ML Engineer
ML EngineerContributions:19 releases, 291 reviews, 45 commits in 7 months
Contributions summary:Ivy's contributions primarily focused on enhancing the `google/flax` repository with support for multi-host checkpointing of JAX GlobalDeviceArrays (GDAs), the preferred array structure for sharded data across multiple hosts. They implemented asynchronous checkpoint saving using thread pools, added and tested the capability to save and restore Flax models containing JAX arrays and other data. The user also contributed significantly to the checkpointing API, introducing new functions such as `save_checkpoint_multiprocess` and modifying core file-handling functions, with detailed error handling and logging.
deep-learningneural-networksneural-networkflaxjax
jax-ml/jax
Jul 2022 - Jul 2022
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:Back-end Developer
Back-end DeveloperContributions:18 reviews, 2 commits, 5 PRs in 2 days
Contributions summary:Ivy contributed to the JAX library by addressing serialization issues within the `jax/experimental/gda_serialization/serialization.py` file, ensuring proper functionality. They also merged changes from the 'google:main' branch into the project, including the `jax/_src/debugger/colab_debugger.py` and `jax/_src/numpy/ufuncs.py` files, indicating involvement in integrating updates and maintaining the codebase. Furthermore, the user implemented improvements to tree utility functions, enhancing the system's tree manipulation capabilities.
pytorchpythonjitautomatic-differentiationgpu
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.