Summary
Alexandra Sidorova is a software engineer specializing in deep learning inference and JIT compilation, with 7 years of experience accelerating models on x86-64, ARM and RISC-V platforms. She spent over five years on Intel’s OpenVINO CPU Plugin and Snippets teams, where she implemented and optimized operators, added dynamic-shape support, and helped JIT-compile subgraphs to speed up transformers, Stable Diffusion and Whisper pipelines. As a GSOC 2024 mentor she led efforts to port and optimize OpenVINO for RISC-V vector extensions, and she has hands-on experience writing JIT kernels with Xbyak and RVV-targeted elementwise kernels. Now at AMD, she contributes to open-source compiler work to maximize inference performance on data-center GPUs, blending low-level assembly-minded optimization with graph-compiler design. A master’s-trained applied mathematician based in Dubai, she pairs deep systems-level expertise with a taste for clever, register-level hacks (as her playful GitHub bio hints).
8 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Information Technology, 5, Bachelor's degree, Information Technology, 5 at State University of Nizhni Novgorod named after N.I. Lobachevsky (UNN)
Master's degree, Applied Mathematics and Informatics, Master's degree, Applied Mathematics and Informatics at Нижегородский Государственный Университет им. Н.И. Лобачевского (ННГУ)