Kumud Bhandari

Compiler Engineer, XLA TPU Optimizations

Houston, Texas, United States
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

👤
Senior
🎓
Top School
Kumud Bhandari is a compiler engineer with 10 years of experience optimizing machine learning workloads for emerging accelerators, currently making Gemini and other generative models run faster on Google TPUs. With a PhD from Rice University and a track record at Facebook AI (PyTorch Glow) and Xilinx, he specializes in backend code generation, IR-to-hardware compiler passes, and heterogeneous runtime systems. He has built modular simulation and verification frameworks for ACAP-class accelerators and shipped production compiler optimizations for novel ML hardware. Comfortable moving between research and production, Kumud blends deep academic rigor with practical engineering to squeeze performance from next-generation silicon. An uncommon asset is his breadth across both hardware-aware simulation and open-source ML compiler projects, enabling rapid delivery of optimized model execution.
code10 years of coding experience
job14 years of employment as a software developer
bookDoctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Rice University
bookBachelors of Science Bachelors of Arts Computers Science Mathematics, Bachelors of Science Bachelors of Arts Computers Science Mathematics at McKendree University
github-logo-circle

Github Skills (8)

hardware9
atlas9
persistent8
persistent-memory8
java7
neural-network6
compiler6
cpp6

Programming languages (1)

C++

Github contributions (3)

github-logo-circle
kumudb/glow

Sep 2019 - Sep 2019

Compiler for Neural Network hardware accelerators
Contributions:2 PRs, 12 pushes in 2 days
hardware-acceleratorscompilerneural-networkacceleratorshardware
HewlettPackard/Atlas

May 2016 - Aug 2016

Atlas: Programming for Persistent Memory
Contributions:47 commits in 3 months
persistent-memorymemorycppatlaspersistent
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