Peter Entschev
Senior System Software Engineer at NVIDIA
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolPeter Entschev is a Senior System Software Engineer with 12 years of experience building high-performance hardware-software systems, currently architecting system software at NVIDIA in Zurich. He combines deep GPU and CUDA expertise with low-level systems work, contributing to cornerstone open-source projects like NumPy, CuPy, Dask, OpenCV and RAPIDS where he improved array protocols, UCX communication, and GPU memory management. His background spans real-time computer vision, CUDA/OpenCL acceleration, and CI/CD/DevOps for GPU-focused ML stacks, reflecting both hands-on optimization and build-system stewardship. Peter’s work often targets subtle correctness and performance issues—implementing array dispatch, stream-aware allocators, and UCX upgrades—rather than headline features. He holds an M.Sc. in Computer Engineering and began his career bridging electronics and software in embedded navigation and institutional IT, which informs his pragmatic hardware-aware approach. Based in Zurich, he thrives at the intersection of systems, HPC and reproducible open-source engineering.
12 years of coding experience
6 years of employment as a software developer
Master of Science (M.Sc.), Computer Engineering, 89%, Master of Science (M.Sc.), Computer Engineering, 89% at Federal Technological University of Paraná
English, German
Stackoverflow
350reputation
5kreached
7answers
0questions
Github Skills (54)
asynchronous10
opencv10
docker10
c-language10
monitor10
set-operations10
pytest10
operation10
python10
async10
testing10
dataframes10
ucx10
memory-management10
cudf10
Programming languages (11)
Github contributions (5)
dask/dask
Feb 2019 - Oct 2024
Parallel computing with task scheduling
Role in this project:
Back-end Developer
Back-end DeveloperContributions:132 reviews, 50 PRs, 888 comments in 5 years 9 months
Contributions summary:Peter primarily contributed to the Dask array library, a component of the Dask parallel computing framework. Their work focused on enhancing the functionality and efficiency of core array operations, specifically related to reductions, sparse array support, and the array function protocol. These contributions included modifications to moment calculations, Cupy test fixes, implementation of the array function protocol, and other optimizations. The user's work suggests an involvement in the core computational logic of the library.
pythonschedulingparallelnumpydask
dask/distributed
Oct 2020 - Nov 2022
A distributed task scheduler for Dask
Role in this project:Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:121 reviews, 34 commits, 61 PRs in 2 years
Contributions summary:Peter primarily contributed to the back-end infrastructure and performance of Dask's distributed task scheduler. They made significant changes to UCX communication layer, including support for UCX 1.10+ and subsequent bug fixes for connection issues and exception handling. Furthermore, the user worked on improving diagnostics and monitoring by integrating with PyNVML for GPU utilization metrics, as well as overall code refactoring and maintenance tasks to ensure the system's stability and reliability.
pythonschedulerdaskdistributed-computingdistributed-systems
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.