ML Framework & Inference Engineer

Systems Engineer focused on ML Infrastructure and High-Performance Computing. I specialize in designing extensible inference engines and optimizing neural network kernels (C++, RVV) using advanced mathematical techniques like Winograd and GEMM.

• Languages: C/C++ 20, Python
• Optimization & Inference: OpenVINO, RVV (RISC-V Vector), Winograd, GEMM, im2col, JIT-compilation
• ML Frameworks: PyTorch, TensorFlow, CatBoost, XGBoost
• Infrastructure: Linux (WSL2), Docker, CMake, CI/CD, GTest

• OpenVINO (YADRO & Intel Industrial Project): Optimized neural network inference for RISC-V architecture. Developed high-performance GEMM and RVV kernels, resulting in a 6x performance increase for specific layers.
• Adept (Personal Project): Developed an Automatic Differentiation Engine. Implemented the Winograd algorithm for 3x3 kernels, achieving 2.25x - 3x speedup during the inference stage.
• Yandex ML Training: Successfully completed Series 1.0, 2.0, and 3.0. Currently focusing on Reinforcement Learning (ML 4.0).

• University: IITMM UNN (Lobachevsky University), CS (FIIT), Department of High-Performance Computing and System Programming.
• Telegram: @idolnot4u
• LinkedIn: linkedin.com/in/strelkovkm
• GitVerse: gitverse.ru/strelkovkm
• Email: strelkovkm96@outlook.com