Liam Fitzpatrick
Engineering Fellow at AMD
Maastricht, Limburg, Netherlands
Summary
🤩
Rockstar🎓
Top SchoolLiam Fitzpatrick is an Engineering Fellow based in Maastricht with over two decades of systems and compiler engineering experience and five years focused on modern ML tooling and architectures. He has progressed from research at Trinity College Dublin to senior engineering and leadership roles at Silexica, Xilinx, and now AMD, blending deep compiler expertise with practical product delivery. Liam contributes to high-profile open-source MLIR work—extending Torch-MLIR with conv2d, maxpool2d refinements and activations like leaky_relu—demonstrating an ability to bridge PyTorch semantics into compiler infrastructures. He’s practiced at scaling parallelization and static-shape reasoning for performance-critical workloads, a skill honed across embedded, FPGA, and compiler toolchains. Colleagues value him for translating research-grade ideas into production-ready implementations and for spotting subtle correctness/performance trade-offs early. Holding a PhD from Trinity College Dublin, he pairs academic rigor with two decades of industrial impact.
5 years of coding experience
31 years of employment as a software developer
PhD, PhD at Trinity College DublinMEng, MEng at University of Limerick
Dutch
Github Skills (12)
mle10
compiler10
pytorch10
compiler-compiler10
python10
mlr10
ml10
data-structures4
algorithm4
algorithms4
data-structure4
tensorflow4
Programming languages (4)
Github contributions (5)
llvm/torch-mlir
Nov 2021 - Mar 2022
The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.
Role in this project:
ML Engineer
ML EngineerContributions:13 reviews, 6 commits, 10 PRs in 4 months
Contributions summary:Liam contributed to the Torch-MLIR project, focusing on extending the support for PyTorch operations within the MLIR ecosystem. They implemented the `leaky_relu` activation function, added bias support to the `aten.conv2d` operation, and refined static shape calculations for `conv2d` and `maxpool2d`. Furthermore, they added support for `constant_pad_nd` and folded operations involving `derefine`.
pytorchmlirtorchcompilerecosystem
ljfitz/mlir-xten
Dec 2021 - Jan 2024
Contributions:87 reviews, 34 PRs, 58 pushes in 2 years 2 months
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.