Justin Luitjens
Director at NVIDIA
Salt Lake City Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolJustin Luitjens is a Director at NVIDIA with 12+ years of engineering and leadership experience focused on GPU-accelerated computing and developer technologies. He progressed from senior engineer to senior manager and now director, guiding teams that bridge low-level CUDA performance work with usable developer tooling. His open-source contributions to high-profile projects like Kaldi and NVIDIA's MatX demonstrate hands-on expertise in CUDA-accelerated feature extraction, batched inference, and efficient C++17 numerical operators. Justin’s work often targets subtle performance bottlenecks—batched matrix copies, memory/stream optimizations, and convolution batching—that materially improve production decoding and ML pipelines. Based in the Salt Lake City area with a PhD in Computer Science from the University of Utah, he combines deep research training with pragmatic engineering leadership. Colleagues rely on him to translate GPU research into robust, production-grade libraries and developer experiences.
12 years of coding experience
14 years of employment as a software developer
The University of Utah
Github Skills (27)
speech-to-text10
c-language10
gpgpu10
matrix10
operation10
tensorrt10
gpu-programming10
hpc10
mat10
tensorflow10
gpu10
cplus10
accelerated-computing10
cuda10
speech-recognition10
Programming languages (6)
Github contributions (5)
NVIDIA/MatX
Jun 2022 - Jan 2023
An efficient C++17 GPU numerical computing library with Python-like syntax
Role in this project:
Back-end Developer
Back-end DeveloperContributions:108 reviews, 77 commits, 113 PRs in 7 months
Contributions summary:Justin primarily contributed to the `nvidia/matx` library, focusing on implementing and improving mathematical operators within the C++17 GPU numerical computing library. Their work included adding new operators like `remap`, `lcollapse`, and `rcollapse` and optimizing existing ones such as `reverse` and `shift`. They also added the `clone` and `slice` operators. Furthermore, the user was involved in enhancing the convolution and correlation functions, including batching and axis support.
cudagpu-programmingnumerical-computingpythoncpp
kaldi-asr/kaldi
Jan 2018 - Jun 2020
kaldi-asr/kaldi is the official location of the Kaldi project.
Role in this project:Back-end Developer / ML Engineer
Back-end Developer / ML EngineerContributions:46 PRs, 169 comments in 2 years 5 months
Contributions summary:Justin made substantial contributions to the Kaldi project, focusing on CUDA-accelerated feature extraction and decoding. They implemented and optimized GPU-based MFCC computation, added batch processing capabilities to neural network inference, and enhanced the CUDA matrix and vector classes. The user also addressed memory copy overhead by implementing batched matrix copy routines, improving the efficiency of the decoding pipeline. Several commits focus on device memory usage and CUDA streams.
cudakaldiasrspeech-to-textkaldi-asr
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.