Elias Stehle
Senior Software Engineer at NVIDIA
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolElias Stehle is a Senior Software Engineer and research scientist specializing in CUDA C++ with over a decade of experience designing scalable, massively parallel GPU algorithms. Based in Zurich and currently at NVIDIA, he focuses on GPU-accelerated data processing and performance-sensitive low-level systems work, including contributions to prominent open-source projects like NVIDIA CUB and RAPIDS cuDF. His research pedigree is strong—several patents and publications in top venues such as VLDB and SIGMOD—and he holds a PhD in Computer Science summa cum laude from TUM. Practically, Elias has improved run-length decoding, namespace handling, and block-level compression/decompression in production-grade GPU libraries, blending academic rigor with engineering pragmatism. Colleagues rely on him for clean refactors that boost performance and testability, and his work often surfaces in benchmarks and test-suite enhancements that make GPU data processing more robust.
6 years of coding experience
Doctor of Philosophy - PhD, Computer Science, summa cum laude (high distinctions), Doctor of Philosophy - PhD, Computer Science, summa cum laude (high distinctions) at Technical University of MunichMaster's Degree, Computer Science, Master's Degree, Computer Science at Technische Universität München
Stackoverflow
103reputation
4kreached
1answer
2questions
Github Skills (16)
compression-algorithm10
cuda10
algorithms10
c-language10
cub10
cprogramming-language10
gpu10
cubit10
c149
data-structure9
c119
cpp9
data-structures9
actionscript6
flash6
Programming languages (4)
Github contributions (5)
NVIDIA/cub
Oct 2020 - Dec 2022
[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl
Role in this project:
Back-end Developer
Back-end DeveloperContributions:145 reviews, 28 commits, 15 PRs in 2 years 2 months
Contributions summary:Elias contributed to the NVIDIA cub library, focusing on run-length decoding and namespace handling. Their work included fixing issues with namespace wrapping, template instantiation, and signedness. The user also refactored the code, incorporating binary search for run-length decoding and adding more efficient implementations. The changes made included improvements to the test suite, adding benchmark functionality.
cxxprimitivesnvidia-hpc-sdknvidiacpp20
rapidsai/cudf
Apr 2021 - Dec 2022
cuDF - GPU DataFrame Library
Role in this project:Back-end Developer
Back-end DeveloperContributions:414 reviews, 15 commits, 28 PRs in 1 year 8 months
Contributions summary:Elias primarily contributed to the `cudf` repository, a GPU DataFrame library. Their work involved removing `nanosleep` calls and refactoring code related to compression and decompression, specifically within the `unsnap.cu`, `gpuinflate.cu`, and `debrotli.cu` files. The commits demonstrate modifications in low-level CUDA code and utilities related to block operations. These changes suggest a focus on optimizing the library's performance.
cudadataframe-librarydata-analysiscppcudf
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.