Ismail Pazarbasi
Staff Machine Learning Engineer at Apple
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolIsmail Pazarbasi is a Staff Machine Learning Engineer with 12 years of experience building high-performance ML inference systems and production infrastructure, now focused on generative AI on Apple Silicon in Palo Alto. He has driven large-scale migrations and optimizations at Meta—delivering 2.5x speedups and halving memory use for recommendation training—while enabling broad deployment of PyTorch 2 runtimes across multiple product teams. His background spans systems, compiler and embedded engineering from Graphcore to Cisco and Opera, and he brings rare compiler expertise evidenced by meaningful contributions to the Clang/LLVM codebase. Comfortable bridging low-level performance work and high-level ML stack design, he combines pragmatic engineering with a track record of unblocking production-scale AI workloads.
12 years of coding experience
20 years of employment as a software developer
Electronics Engineering, Electronics Engineering at Fatih UniversityElectronics, Electronics at Tekirdağ Teknik Lisesi
English, Norwegian, Turkish
Github Skills (10)
compiler10
compiler-compiler10
c-language10
cprogramming-language10
memory-management9
static-analysis8
debug8
debugging8
performance-analysis7
performance-monitor7
Programming languages (4)
Github contributions (5)
llvm-mirror/clang
Oct 2013 - Sep 2016
Mirror kept for legacy. Moved to https://github.com/llvm/llvm-project
Role in this project:
Back-end Developer & Compiler Engineer
Back-end Developer & Compiler EngineerContributions:16 commits in 2 years 11 months
Contributions summary:Ismail's contributions primarily involved fixing compiler-related issues, particularly within the Clang codebase. They addressed bugs related to header guard mismatches, string literal conversions, and initialization sequences, ensuring the compiler's correct behavior. Several commits focused on improving the handling of C++ features, such as overload resolution and user-defined conversions, demonstrating a strong understanding of the language's intricacies. Furthermore, the user addressed potential memory safety issues by correcting initialization problems in standard conversion sequences.
keptwindowsllvmcc-plus-plus
ipazarbasi/compiler-rt
Feb 2015 - Nov 2018
Contributions:9 pushes, 1 branch in 3 years 9 months
git-repositoryhourlycompilerupdatedcompiler-rt
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.