Junmin Hao - Quality Lead - Google Workspace AI Platform

Junmin Hao

Quality Lead - Google Workspace AI Platform

Sunnyvale, California, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Junmin Hao is a seasoned software and quality leader with 24+ years of experience building and scaling machine learning, distributed systems, and ML infrastructure at Google, AWS, and Microsoft. Currently leading quality for the Google Workspace AI Platform, he focuses on taming LLMs into reliable, production-ready capabilities and has driven multiple technical lead roles reinventing Google Assistant with LLMs. His hands-on background spans embedded systems, large-scale ad and search backends, and ML accelerator enablement—helping customers adopt AWS Trainium/Inferentia and contributing backend improvements to the widely used pytorch/xla project to enable TPU distributed primitives. Known for blending engineering rigor with product-level quality ownership, Junmin has a track record of reducing resource footprints and shipping end-to-end ML pipelines. Based in Sunnyvale, he pairs a physics foundation from Peking University with continuous learning in AI and CS, bringing both systems-level depth and pragmatic leadership to complex ML productionization.

4 years of coding experience

24 years of employment as a software developer

Bachelor Physics, Bachelor Physics at Peking University

None Artificial Intelligence, None Artificial Intelligence at Udacity

No degree Computer Science, No degree Computer Science at Stanford University

Github Skills (10)

xla10

compiler10

pytorch10

compiler-compiler10

distributed-training10

c-language10

backend10

cprogramming-language10

back-end-development10

python8

Programming languages (3)

C++CPython

Github contributions (5)

pytorch/xla

Jan 2022 - Aug 2022

Enabling PyTorch on XLA Devices (e.g. Google TPU)

Role in this project:

Backend Developer

Contributions:65 reviews, 5 commits, 7 PRs in 6 months

Contributions summary:Junmin primarily contributed to the implementation and enhancement of XLA support within the PyTorch ecosystem. Their work included implementing the `all_gather` primitive using native XLA operations and adding XLA backend support for the `torch.distributed` module, enabling collective communication operations. Further contributions involved incorporating `send` and `recv` functionality to enable inter-process communication. These changes aimed to improve the integration of PyTorch with XLA devices, particularly for distributed training.

pytorchxladeep-learningtpucompiler

hjm-aws/xla

Jan 2022 - Aug 2022

Enabling PyTorch on Google TPU

Contributions:4 PRs, 57 pushes, 8 branches in 7 months

pytorchtpu

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.

Request Free Trial