Suraj Sudhir
Senior Principal Software Engineer at Arm
California, United States
Summary
🤩
Rockstar🎓
Top SchoolSuraj Sudhir is a senior principal software engineer based in California with deep expertise in ML compilers, hardware-aware performance modeling, and end-to-end compiler stacks. He has driven PyTorch->TOSA integration and helped create the TOSA MLIR dialect while contributing backend support to the high-profile llvm/torch-mlir project, implementing ops like conv2d, matmul and aten.embedding. At Cruise he led design and delivery of a full ML compiler pipeline for heterogeneous vehicle platforms, coordinating across models, runtimes and hardware architects; at Arm he builds training and inference infrastructure spanning PyTorch, Triton and TOSA. His background blends chip-level performance validation and simulator development from his Sun/Oracle years with practical compiler and quantization numerics experience, enabling productive hardware–software co-design. Colleagues rely on him to translate complex hardware constraints into robust compiler implementations and deployment tooling.
5 years of coding experience
24 years of employment as a software developer
Kendriya Vidyalaya SangathanB.E. Computer Science and Engineering, B.E. Computer Science and Engineering at College of Engineering Guindy, ChennaiMS Computer Engineering, MS Computer Engineering at Carnegie Mellon University
English, c/c++, unix shell, perl, python
Github Skills (8)
pytorch10
togaf10
lsa10
tosca10
mlr10
compiler-compiler9
compiler9
python8
Programming languages (3)
Github contributions (5)
llvm/torch-mlir
Oct 2021 - Jul 2022
The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:115 reviews, 17 commits, 86 PRs in 8 months
Contributions summary:Suraj primarily contributed to the Torch-MLIR project by implementing support for various TOSA operations, including sigmoid, tanh, argmax, squeeze, matmul, and conv2d. They refactored existing code to use templates and added more utility functions for TOSA legalization. The user also made modifications related to the RefineTypes pass and implemented support for the aten.embedding operation, enhancing the project's capabilities in supporting PyTorch models.
pytorchmlirtorchcompilerecosystem
sjarus/torch-mlir
Nov 2021 - Sep 2024
The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.
Contributions:78 pushes, 36 branches in 2 years 11 months
pytorchmlirdeep-learningtorchecosystem
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.