Peter Mcaughan
Senior System Software Engineer, AI at NVIDIA
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPeter Mcaughan is a Senior System Software Engineer specializing in AI with eight years of experience building high-performance ML inference and cloud systems from research to production. Based in San Francisco, he contributed significant optimizations to the widely used ONNX Runtime—adding memory configuration options, Whisper model export/integration, and CUDA attention kernels—before joining NVIDIA to work on system-level AI software. His background blends academic research in ubiquitous computing and sign-language recognition with hands-on Azure Compute and cloud platform engineering, giving him a pragmatic research-to-product perspective. Known for digging into performance and memory bottlenecks, he brings a knack for making large models run faster and leaner in real-world environments.
8 years of coding experience
6 years of employment as a software developer
Master's degree, Computer Science, Master's degree, Computer Science at Georgia Institute of TechnologyBachelor’s Degree, Computer Engineering, Bachelor’s Degree, Computer Engineering at Texas A&M UniversityUniversity of São Paulo
English
Github Skills (11)
cuda10
machine-learning10
deep-learning10
onnx10
python9
pytorch9
c-language8
ai-framework8
neural-network8
hardware-acceleration8
cprogramming-language8
Programming languages (4)
Github contributions (5)
microsoft/onnxruntime
Sep 2022 - Oct 2022
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
Role in this project:
ML Engineer
ML EngineerContributions:28 reviews, 2 commits, 29 PRs in 1 month
Contributions summary:Peter's contributions primarily focused on enhancing the ONNX Runtime, specifically related to integrating and optimizing models for machine learning inference and training. Their work involved adding support for various memory configuration options and implementing features for the Whisper model, including export scripts and integration with the BeamSearch operation. They also addressed performance bottlenecks and memory usage issues during the model conversion process. Further contributions included CUDA kernel implementations and optimization efforts for the attention layer within the CUDA environment.
runtimetrainingtensorflowai-frameworkaccelerator
petermcaughan/pmcaughanSite
Jul 2017 - Jun 2019
Source code for my personal website
Contributions:36 pushes, 1 branch in 1 year 11 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.