Martin Marenz
Senior AI Expert at yasp
Leipzig, Saxony, Germany
Summary
🤩
Rockstar🎓
Top SchoolMartin Marenz is a Senior AI Expert with a PhD in physics and over a decade of experience building high-performance systems across HPC, embedded firmware, and machine learning. He has run and optimized Linux HPC clusters, developed C/C++ firmware for telematics, and driven production ML at companies from Webfleet to Meta and AWS. Recently focused on agentic AI compilation at yasp, he brings kernel- and hypervisor-level performance insight combined with data-science rigor. An active contributor to GPU data tooling, he improved cuDF benchmarking metrics to better measure throughput and null-handling behavior—reflecting a practical focus on measurable performance gains. Based in Leipzig, he blends theoretical physics training with hands-on systems engineering to solve computation- and latency-critical problems.
10 years of coding experience
7 years of employment as a software developer
Doktor (Ph.D.) Physik, Doktor (Ph.D.) Physik at Leipzig UniversityAbitur, Abitur at Gymnasium Bertolt Brecht
German, English, French
Stackoverflow
2,802reputation
216kreached
8answers
7questions
Github Skills (18)
benchmark10
c-language10
cudf10
benchmarking10
gpu10
performance-optimization10
cuda10
cpp10
cprogramming-language10
data-science9
data-analysis9
tango6
parsing6
aio6
autocomplete6
Programming languages (6)
Github contributions (5)
rapidsai/cudf
Aug 2023 - Oct 2023
cuDF - GPU DataFrame Library
Role in this project:
Performance Engineer
Performance EngineerContributions:6 reviews, 8 PRs, 16 comments in 1 month
Contributions summary:Martin primarily focused on enhancing the performance analysis of the cuDF benchmarks. Their contributions involved adding `bytes_per_second` metrics to existing benchmarks such as `APPLY_BOOLEAN_MASK`, `compiled_binaryop`, `copy_if_else`, `hash_partition`, `distinct_count`, `transpose`, and `shift`. They also enabled fractional null probabilities in the hashing benchmark. This work improves the ability to measure and optimize the performance of cuDF operations.
cudadataframe-librarydata-analysiscppcudf
Blonck/rl
Apr 2023 - Aug 2023
A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.
Contributions:131 pushes, 23 branches in 4 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.