Pekka Jääskeläinen
Tampere, Pirkanmaa, Finland
Summary
🤩
RockstarPekka Jääskeläinen is a senior compiler and heterogeneous computing engineer with 18 years of experience focused on code generation, customized processors, and OpenCL-enabled backends. Based in Tampere, Finland, he has made notable open-source contributions to major projects like Clang (LLVM) and Glow, improving OpenCL support, floating-point correctness, address-space handling, and performance-critical kernel specializations. His work spans modernizing build systems (pocl/CMake), kernel compiler tweaks, and performance engineering such as operation fusion and disk caching for accelerator workflows. Comfortable at the intersection of low-level compiler internals and hardware-aware optimizations, he brings practical improvements that directly boost runtime efficiency and portability. An often-overlooked strength is his consistent attention to build robustness across LLVM versions, ensuring long-term maintainability of complex toolchains.
18 years of coding experience
Stackoverflow
111reputation
1kreached
5answers
0questions
Github Skills (23)
opencl10
c-language10
llvm10
build-system10
cmake10
c1110
c1710
compiler-compiler10
cplus10
compile10
neural-network10
compiler-optimization10
performance-tuning10
network10
compiler10
Programming languages (9)
Github contributions (5)
pocl/pocl
Oct 2011 - Jan 2023
pocl - Portable Computing Language
Role in this project:
Backend Developer
Backend DeveloperContributions:1 release, 510 reviews, 1968 commits in 11 years 5 months
Contributions summary:Pekka's commits primarily focused on adapting the build environment and codebase of the PoCL project to use modern build systems like CMake. They removed dependencies on older tools, such as Autotools, in the build scripts. The commits included modifications to the program's source code for handling command-line arguments. The commits showed also changes related to printf function calls, suggesting the user worked on the kernel compiler implementation.
heterogeneous-parallel-programmingcompilerportablecomputingprogramming-language
pytorch/glow
Mar 2019 - Mar 2020
Compiler for Neural Network hardware accelerators
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:17 commits, 19 PRs, 78 comments in 11 months
Contributions summary:Pekka primarily focused on optimizing and improving the performance of the Glow compiler for neural network hardware accelerators. Their contributions include addressing build issues with different LLVM versions, and adding specialized OpenCL kernels to improve efficiency. Significant work involved optimizing OpenCL kernels through specialization and constant propagation, demonstrating a strong understanding of compiler optimization techniques. The user also worked on fusing operations and implementing performance-related features like disk caching for OpenCL programs, leading to significant speedups.
hardware-acceleratorscompilerneural-networkacceleratorshardware
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.