Philip Lassen
Staff Software Engineer at NVIDIA
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPhilip Lassen is a Staff Software Engineer based in San Francisco with eight years of experience specializing in compilers and MLIR-based tooling. As co-chair of the ONNX Compiler SIG and a contributor to onnx-mlir, he has deep expertise in shape inference and opset evolution for production-grade model lowering. His backend work spans niche languages and targets—improving Futhark’s interpreter and adding WebAssembly support—demonstrating both systems-level rigor and attention to developer experience. Philip has progressed from research and teaching roles to industry compiler engineering at Groq and NVIDIA, blending a mathematics foundation with an MS in Computer Science. He prefers rolling up his sleeves on tricky verification and codegen problems, and his GitHub work shows a knack for pragmatic refactors that reduce bugs and improve reuse.
8 years of coding experience
5 years of employment as a software developer
Bachelor of Science - BS Mathematics, Bachelor of Science - BS Mathematics at University of WashingtonMaster of Science - MS Computer Science, Master of Science - MS Computer Science at Københavns Universitet - University of Copenhagen
Github Skills (23)
c-language10
python10
inference10
onnx10
mlr10
compiler-compiler10
tensorflow10
computer-engineering10
compiler10
tensor10
webassembly10
cprogramming-language10
shapes10
back-end-development9
os-development9
Programming languages (13)
Github contributions (5)
onnx/onnx-mlir
Apr 2022 - Jan 2023
Representation and Reference Lowering of ONNX Models in MLIR Compiler Infrastructure
Role in this project:
Back-end Developer
Back-end DeveloperContributions:118 reviews, 28 commits, 109 PRs in 8 months
Contributions summary:Philip primarily focused on enhancing the ONNX-MLIR project by implementing shape inference capabilities for various ONNX operators. They added and improved shape inference for operators like `Celu`, `EyeLike`, `Upsample`, and `MatMulInteger`, indicating a strong understanding of the ONNX standard and the underlying MLIR infrastructure. Furthermore, the user contributed to the project by integrating opset 15 and 16 changes and fixing verification issues for operations. This includes refactoring and restructuring existing shape inference functions using modern C++ approaches to promote code reuse.
pytorchrepresentationdeep-learningmlironnx-models
diku-dk/futhark
Aug 2020 - Sep 2021
:boom::computer::boom: A data-parallel functional programming language
Role in this project:Back-end Developer
Back-end DeveloperContributions:17 reviews, 254 commits, 35 PRs in 1 year 1 month
Contributions summary:Philip primarily contributed to the backend of the Futhark compiler, specifically improving error messages, adding checks for dimension and type mismatches in the Python backend, and fixing a division-by-zero error in the interpreter. They also addressed code quality by removing dead code and fixing a typo. Furthermore, the user added support for the WebAssembly backend, including necessary code generation and testing infrastructure.
cudafutharkgpu-programminggpu-accelerationfunctional-programming
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.