Luke Yeager
Software Engineer at NVIDIA
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolLuke Yeager is a seasoned software engineer with 14 years of experience building and operating high-performance systems for deep learning, HPC, and autonomous vehicle data infrastructure. Based in Austin, he has held multiple roles at NVIDIA and Tesla, contributing to projects from DIGITS and Caffe/Caffe2 to datacenter systems and Autopilot data pipelines. He blends backend development and DevOps expertise—improving build/test infrastructure, container tooling, and cluster automation—and has hands-on experience integrating Slurm, Kubernetes, and MPI in production-like environments. His open-source contributions include practical enhancements to NVIDIA enroot and deepops, plus low-level plugins for collectd, demonstrating a knack for making complex tooling more reliable and user-friendly. Notably, he optimizes developer workflows (e.g., removing unnecessary CUDA dependencies to speed builds) and surfaces hard-to-find bugs in high-performance codepaths. He pairs a Computer Engineering degree from Texas A&M with systems coursework at Stanford, reflecting both formal training and continuous learning.
14 years of coding experience
4 years of employment as a software developer
Bachelor's Degree Computer Engineering, Bachelor's Degree Computer Engineering at Texas A&M UniversityCourses in Systems Software, Courses in Systems Software at Stanford University
Stackoverflow
1,410reputation
350kreached
53answers
6questions
Github Skills (56)
infiniband10
hdf10
kubernetes10
c-language10
caffe10
slurm10
python10
scripting10
information-gathering10
testing10
machine-learning10
linux10
bash10
hpc10
cmake10
Programming languages (20)
Github contributions (5)
NVIDIA/DIGITS
Mar 2015 - Jul 2017
Deep Learning GPU Training System
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:18 releases, 1059 commits, 847 PRs in 2 years 4 months
Contributions summary:Luke contributed to the NVIDIA Digits project by working on the backend, including code modifications for network training processes, bug detection tests, and data management tasks. The user also introduced features related to job management, such as adding a download link for Caffe output and an option to reuse previous models for fine-tuning. Furthermore, the user also made adjustments to improve build process and to ensure better testing practices through the addition of tests and by addressing issues related to the project's configuration and environment setup.
deep-learninggputorchmachine-learningtraining
NVIDIA/enroot
May 2019 - Feb 2021
A simple yet powerful tool to turn traditional container/OS images into unprivileged sandboxes.
Role in this project:DevOps Engineer
DevOps EngineerContributions:3 reviews, 9 commits, 11 PRs in 1 year 9 months
Contributions summary:Luke contributed to the project by addressing issues related to the integration of Slurm with PMIx, ensuring correct detection of MPI types. They implemented enhancements for Docker image handling and provided options to set timeouts. The user also fixed bugs related to argument parsing and directory permissions within the codebase. Overall, the user's work focuses on improving the tool's usability and integration with HPC environments.
containersdocker-imagetraditionalunprivilegedsimple-yet-powerful
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.