Andrey Borzunov

I’m a C++ developer and applied mathematician with six years of experience building high-performance numerical software for both industry and research. My work lives at the intersection of rigorous mathematics and low-level performance engineering, where I can own problems end-to-end—from formulation to profiled, production-ready implementation.

At Severstal, I develop and optimize numerical schemes for physical modeling, including nonlinear heat transfer and steel hot-rolling. I focus on pushing performance through CUDA, SIMD (AVX/AVX2/AVX512), MPI, custom allocators, cache-friendly memory layouts, and careful profiling.

Earlier, as a researcher at MSU, I worked on numerical methods for ODEs, PDEs, nonlinear PDEs, optimization, and inverse/ill-posed problems. I also built massively parallel software for the Lomonosov-2 supercomputer.

I value correctness and speed in equal measure: verified numerical schemes, reproducible builds, and CI/CD that keeps both intact. I’m the author of several peer-reviewed papers in international journals (Elsevier, MDPI, Pleiades), and I’m open to challenging problems in high-performance computing, numerical modeling, optimization, and performance-critical C++.