Luke Hutton
Staff Compiler Engineer at Arm
Cambridge, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolLuke Hutton is a Staff Compiler Engineer based in Cambridge with eight years of experience accelerating machine learning workloads through compiler and backend engineering at Arm. He specializes in ML compilers, MLIR and the TOSA dialect, with notable open-source contributions to Apache TVM, LLVM, and TensorFlow that optimize models for Arm Ethos NPUs and SME vector extensions. His work spans code generation, verifier improvements and operation legalization—translating TensorFlow and TFLite ops into efficient TOSA primitives and fixing subtle correctness issues like conv2d/verifier bugs and integer division edge cases. Having progressed from intern to staff engineer at Arm, he blends low-level systems expertise with practical deployment experience on edge devices. An active GitHub maintainer, he contributes to widely used projects in the deep learning compiler ecosystem, bringing both architect-level design and hands-on implementation to production toolchains.
8 years of coding experience
7 years of employment as a software developer
Arnold Hill AcademyBSc Computer Science with Industrial Placement, Computer Science, BSc Computer Science with Industrial Placement, Computer Science at University of Leeds
Github Skills (22)
c-language10
tosca10
operation10
python10
llvm10
tensorrt10
machine-learning10
lsa10
mlr10
intermediate10
togaf10
code-generation10
multilevel10
tensorflow10
arm10
Programming languages (5)
Github contributions (5)
apache/tvm
Oct 2019 - Dec 2022
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
ML Engineer & Software Architect
ML Engineer & Software ArchitectContributions:625 reviews, 116 commits, 328 PRs in 3 years 2 months
Contributions summary:Luke focused on improving the TVM compiler stack for deep learning, with a strong emphasis on optimizing for Arm(R) Ethos(TM)-U NPU devices. Their commits involved significant changes to the code generation for SME (Scalable Matrix Extension) intrinsics and buffer load/store predication. This included contributions to the conversion of operations like multiply and add to depthwise convolutions, and fixing output issues with specific operations.
metalvulkancompilertensoropencl
llvm/llvm-project
Jan 2023 - Jan 2023
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Role in this project:Back-end Developer
Back-end DeveloperContributions:118 reviews, 1 commit, 20 PRs in 1 day
Contributions summary:Luke primarily focused on the TOSA (Tensor Operator Set Architecture) dialect within the LLVM project. Their contributions involved enhancing the validation process, including stricter checks for element types, handling dialect-specific operations, and adding verifiers for FFT2d and RFFT2d. The user also fixed bugs related to integer division by zero and improved the conv2d verifier, contributing to the overall correctness and robustness of the TOSA implementation.
compilerstechnologiesclangsubmittoolchain
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.