Douglas Lehr
Principal Engineer On MLSE Frameworks Team at AMD
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolDouglas Lehr is a Principal Engineer based in Austin with a decade-plus track record of bringing deep hardware knowledge to production ML frameworks, currently integrating next-generation AMD hardware into upstream deep learning stacks. He combines kernel- and firmware-level experience with applied deep learning expertise—contributions to PyTorch enabling ROCm support, CUDA-graph functionality, and Triton/MLIR integration show his influence on a widely used open-source framework. Previous roles at IBM span kernel development, PowerAI enablement, systems architecture, and leading design-driven deep learning applications for manufacturing, reflecting both low-level systems craftsmanship and product-focused delivery. Comfortable across assembly, kernel code, cloud-native deployment, and ML framework internals, he excels at translating chip architecture into usable software primitives. Colleagues would note his appetite for learning and a rare mix of academic rigor (MS in CS) with pragmatic system-building.
4 years of coding experience
16 years of employment as a software developer
M.S., Computer Science, M.S., Computer Science at Northern Illinois University
Github Skills (12)
pytorch10
machine-learning10
roc10
gpu10
python10
triton9
cuda9
deep-learning8
deeplearning-ai8
autograd7
tensor7
build-automation6
Programming languages (3)
Github contributions (5)
pytorch/pytorch
Oct 2021 - Aug 2022
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:4 reviews, 9 commits, 9 PRs in 9 months
Contributions summary:Douglas primarily contributed to adding and improving ROCm (AMD GPU) support within the PyTorch framework, focusing on graph optimizations and Triton integration. Their work included enabling CUDA graph functionality for ROCm, adding support for the pytorch-triton-mlir backend, and modifying build scripts for Triton wheels. The user also addressed compatibility issues and debug mode settings for ROCm, enhancing the framework's usability on AMD GPUs.
pythongpu-accelerationdeep-learninggpunumpy
dllehr-amd/triton
Feb 2023 - Aug 2023
Development repository for the Triton language and compiler
Contributions:2 PRs, 37 pushes, 18 branches in 5 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.