Jiewen Tan
Staff Research Engineer at Google DeepMind
Sunnyvale, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJiewen Tan is a Staff Research Engineer with 11 years of experience building low-level systems and scalable ML infrastructure, currently driving GenAI work at Google DeepMind. He has led framework and distributed modeling efforts across Google and Meta, including core contributions to PyTorch, PyTorch/XLA, and the Lazy Tensor backend that removed legacy view/aliasing and improved TPU support. Earlier at Apple he was a top WebKit contributor who led Web Authentication, Web Cryptography, and Face/Touch ID integrations for the web, representing Apple in W3C and FIDO standards. Comfortable toggling between production-grade browser security and high-performance ML kernels, he combines deep systems expertise with standards-level product impact. Notably, his open-source work touches foundational projects like pytorch/pytorch, reflecting both debugging and architectural refactors that enable large-model training on accelerators.
11 years of coding experience
9 years of employment as a software developer
The University of Hong Kong (HKU)University of California, Los Angeles
Github Skills (17)
pytorch10
c-language10
operation10
back-end-development10
python10
tensorrt10
deep-learning10
tensorflow10
xla10
tensor10
cprogramming-language10
machine-learning9
autograd9
compiler8
compiler-compiler8
Programming languages (7)
Github contributions (5)
pytorch/xla
Sep 2022 - Jan 2023
Enabling PyTorch on XLA Devices (e.g. Google TPU)
Role in this project:
Back-end Developer
Back-end DeveloperContributions:863 reviews, 400 commits, 379 PRs in 4 months
Contributions summary:Jiewen contributed to the PyTorch/XLA repository, which focuses on enabling PyTorch on XLA devices. The user's work centered on implementing and supporting the `Tensor.is_alias_of` functionality within the XLA backend. This involved adding support for view tensors, making expand operations compatible, and improving the DistributedDataParallel (DDP) implementation for correctness and larger model testing. These changes suggest a focus on improving the core functionality and performance of PyTorch on XLA devices.
pytorchxladeep-learningtpucompiler
pytorch/pytorch
Sep 2021 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:Back-end Developer
Back-end DeveloperContributions:402 reviews, 143 commits, 224 PRs in 1 year 4 months
Contributions summary:Jiewen primarily contributed to the Lazy Tensor Core (LTC) within the PyTorch repository, focusing on refactoring and removing legacy components. Their work involved eliminating deprecated view and aliasing infrastructure in favor of functionalization, and removing non-native view operations. They also modified core files related to tensor operations and graph execution, demonstrating expertise in the internal workings of the PyTorch framework. Further contributions included the restoration of a debugging utility and refinements to metrics, ensuring maintainability and debugging capabilities.
pythongpu-accelerationdeep-learninggpunumpy
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.