Egor Churaev
Sr. Software Engineer at Deelvin Solutions Inc.
Russia
Summary
🤩
Rockstar🎓
Top SchoolEgor Churaev is a senior software engineer with 11 years of experience specializing in back-end systems, compilers and deep learning runtimes, currently at Deelvin Solutions. He holds a PhD in Mathematics and Computer Science from HSE and is an active committer on Apache TVM, contributing cross-platform runtime fixes and ONNX frontend operators. Egor has a strong Intel pedigree, where he worked extensively on OpenCL/Clang and OpenVINO, implementing new deep-learning ops and low-level optimizations (including int8 fsv16 kernels) that improve inference performance. His open-source work spans SPIR-V/LLVM translation and compiler internals, reflecting deep expertise in IR, address-space handling and codegen. Notably, he blends academic rigor with production-grade engineering—tracking down subtle memory leaks and RPC platform quirks while adding standardized operator support for major ML toolchains.
11 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD, Mathematics and Computer Science, Doctor of Philosophy - PhD, Mathematics and Computer Science at Higher School of EconomicsMaster's degree, Applied Mathematics, 4.92, Master's degree, Applied Mathematics, 4.92 at Nizhniy Novgorod State Technical University named after R.Y. Alekseev (NSTU)
Russian, English, German
Stackoverflow
3reputation
103reached
1answer
1question
Github Skills (31)
spirv10
opencl10
c-language10
llvm10
inference10
onnx10
c1110
openvino10
c1710
compiler-design10
compiler-compiler10
deep-learning10
ai10
cplus10
compiler10
Programming languages (10)
Github contributions (5)
apache/tvm
Dec 2020 - Jan 2023
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
Back-end Developer & System Architect
Back-end Developer & System ArchitectContributions:512 reviews, 56 commits, 149 PRs in 2 years 1 month
Contributions summary:Egor primarily focused on enhancing the Apache TVM compiler stack by addressing issues related to RPC communication on macOS, specifically preferring IPv4 over IPv6 in RPC connections. They implemented the CumSum operator within the ONNX frontend, including support for exclusive and reverse attributes, along with comprehensive testing and integration. Furthermore, the user addressed memory leaks and other issues in the Metal runtime and added features and fixes to enhance the functionality and stability of the TVM stack on various platforms.
metalvulkancompilertensoropencl
openvinotoolkit/openvino
May 2020 - Nov 2020
OpenVINO™ is an open-source toolkit for optimizing and deploying AI inference
Role in this project:Backend Developer
Backend DeveloperContributions:12 reviews, 11 commits, 13 PRs in 5 months
Contributions summary:Egor implemented new operations for the OpenVINO toolkit related to Deep Learning inference. Their work included adding the CumSum, EmbeddingBag, NormalizeL2, ExtractImagePatches, and Interpolate-4 operations. Furthermore, the user enhanced existing kernels, such as LRN int8 fsv16 optimizations and the fsv16 i8 pooling kernel, indicating a focus on performance optimization. The changes also involved fixing issues related to device release, reshape operations, and linear_onnx Interpolate selection, demonstrating a strong understanding of the underlying framework.
inference-enginepytorchmodel-optimizerdeep-learninggpu
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.