Patrice Castonguay
Principal Engineer, LLM Inference at NVIDIA
Greater Montreal Metropolitan Area Canada
Summary
🤩
Rockstar🎓
Top SchoolPatrice Castonguay is a Principal Engineer at NVIDIA specializing in accelerating LLM inference, with a deep background in high-performance computing and numerical methods from a PhD in Aeronautics and Astronautics at Stanford. Over a 15+ year technical career spanning research and industry, he has built GPU-accelerated solvers, led CFD algorithm development at Bombardier, and authored performance-critical libraries and patents for sparse linear systems. He combines low-level optimization (CUDA, MPI, C/C++) with algorithmic innovation to scale workloads on modern Intel and NVIDIA architectures. Known for translating academic research into production-grade software, he often bridges multidisciplinary teams to deliver measurable speedups and novel parallel algorithms. Based in Greater Montreal, he brings an uncommon blend of aerospace CFD expertise and practical ML/LLM systems engineering to inference performance challenges.
6 years of coding experience
9 years of employment as a software developer
Bachelor Mechanical Engineering - Aeronautical Concentration, Bachelor Mechanical Engineering - Aeronautical Concentration at McGill UniversityPh. D. Aeronautics and Astronautics, Ph. D. Aeronautics and Astronautics at Stanford University
French, English
Github Skills (21)
pytorch10
runtimes10
python10
tensorrt10
inference10
large-language-models10
llm10
nvidia10
cuda10
cpp10
backend10
command-line7
grpc6
cli6
protobuf6
Programming languages (4)
Github contributions (5)
pcastonguay/TensorRT-LLM
Mar 2025 - May 2026
TensorRT-LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and build TensorRT engines that contain state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT-LLM also contains components to create Python and C++ runtimes that execute those TensorRT engines.
Contributions:147 pushes, 44 branches in 1 year 1 month
cppinferencelarge-language-modelsllmnvidia
pcastonguay/cpp-clients
Mar 2022 - Aug 2022
Sample C++ command-line Riva clients.
Contributions:16 pushes, 9 branches in 5 months
clientscpprivacli
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.