Joe Shajrawi
Senior Engineering Manager at CoreWeave
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolJoe Shajrawi is a Senior Engineering Manager based in Austin with 9 years of industry experience leading teams that build high-performance compilers and systems software. He has driven compiler and hardware/software co-design efforts at Meta (notably on first-generation silicon for recommendation systems) and has led engineering at AMD and CoreWeave, combining leadership with deep hands-on expertise in LLVM, OpenCL, and performance optimization. His background spans research and product engineering—from dynamic binary optimization and cybersecurity research at IBM to Swift compiler work at Apple—giving him a rare blend of academic rigor and production delivery. An active contributor to the PyTorch Glow project, he implemented operators and backend fixes that improved neural network accelerator support, reflecting his practical impact on ML compiler toolchains. A US citizen with an M.Sc. in Computer Science from the University of Haifa, he excels at translating low-level systems complexity into scalable, deployable solutions.
9 years of coding experience
19 years of employment as a software developer
M.Sc. Computer science, M.Sc. Computer science at University of Haifa
Github Skills (9)
neural-network10
network10
c-language10
cprogramming-language10
compile10
machine-learning9
caffe9
opencl7
python4
Programming languages (3)
Github contributions (5)
pytorch/glow
Jul 2019 - Jul 2020
Compiler for Neural Network hardware accelerators
Role in this project:
Back-end Developer
Back-end DeveloperContributions:2 reviews, 22 commits, 22 PRs in 1 year
Contributions summary:Joe primarily contributed to the `glow` repository, which is a compiler for Neural Network hardware accelerators, by implementing new operators and fixing bugs related to existing functionalities. The user added support for the Bucketize operator, implementing the Caffe2 Bucketize functionality, and resolved issues related to VectorFloat hashing. Furthermore, the user addressed a bug in the foldChannelShuffle optimization and added support for several operators such as LengthsSum, LengthsRangeFill, and BatchOneHot on the Habana Backend, enhancing the backend support.
hardware-acceleratorscompilerneural-networkacceleratorshardware
shajrawi/glow
Jul 2019 - Jul 2020
Compiler for Neural Network hardware accelerators
Contributions:194 pushes, 39 branches, 4 comments in 1 year
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.