Andrey Turetskiy
Principal Software Engineer at Cadence Design Systems
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolAndrey Turetskiy is a Principal Software Engineer based in the San Francisco Bay Area with 13 years of experience building and optimizing compilers and low-level libraries. He specializes in x86 performance and code-size tuning, with a proven track record at Intel and Cadence of squeezing measurable percent-level gains from toolchains and runtimes. Andrey has deep compiler-internals expertise—contributing fixes and architecture-specific support to Clang/LLVM and to the Glow neural-network compiler back end—demonstrating comfort across both system-level and ML-accelerator toolchains. He designed MCU compilation infrastructure, enabled MSVC/Intel target features, and shipped extensive AVX‑512 and intrinsic support in GCC, showing a knack for CPU-specific instruction work that’s rarely visible to end users. Colleagues rely on him for stabilizing performance testing, automating backend validation, and pragmatic designs that bridge prototypes to production. Trained at MIPT, he combines rigorous academic grounding with hands-on open-source impact on widely used compiler projects.
13 years of coding experience
6 years of employment as a software developer
Master’s Degree Computer Systems and Networks, Master’s Degree Computer Systems and Networks at Moscow Institute of Physics and Technology (State University) (MIPT)
English, Russian
Github Skills (23)
unit-testing10
c-language10
back-end-development10
c1110
onnx10
x8610
c1710
compiler-compiler10
compile10
neural-network10
network10
compiler10
cprogramming-language10
operation9
test-framework9
Programming languages (3)
Github contributions (5)
pytorch/glow
Jul 2018 - Jul 2021
Compiler for Neural Network hardware accelerators
Role in this project:
Back-end Developer & Test Automation Engineer
Back-end Developer & Test Automation EngineerContributions:16 reviews, 57 commits, 57 PRs in 3 years
Contributions summary:Andrey primarily contributed to the Glow compiler for neural network hardware accelerators by adding comments, fixing issues related to S8 tensor generation, and improving the CPU backend. They addressed bugs in the unit tests and improved the functionality of CPU-specific optimizations and the Caffe2 importer. Additionally, the user added support for new types (Int16QTy) and included features like the ONNX Clip operator, PPM image support, and the loop operator, indicating a focus on expanding the library's capabilities and improving testing.
hardware-acceleratorscompilerneural-networkacceleratorshardware
llvm-mirror/clang
Feb 2016 - Jun 2016
Mirror kept for legacy. Moved to https://github.com/llvm/llvm-project
Role in this project:Backend Developer
Backend DeveloperContributions:16 commits in 4 months
Contributions summary:Andrey primarily focused on low-level compiler and system-level tasks within the Clang compiler. Their contributions involved fixing stack alignment issues for the Intel MCU target, enabling the `--rtlib` option for the MSVC target, and adding support for the Lakemont processor. They also added and modified x86 target features and options, demonstrating a deep understanding of compiler internals and target architecture specifics.
keptwindowsllvmcc-plus-plus
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.