Andrey Krisanov is a Staff Software Engineer at Severstal, building AI infrastructure and ML systems.
He leads the architecture and hands-on development of DaVinci, a shared GenAI platform running production workloads across multiple data centers on Kubernetes and NVIDIA GPUs. His work covers platform architecture, GPU infrastructure and capacity planning, reliability, observability, and model lifecycle in production.
A large part of his current work is LLM inference: model serving with vLLM, traffic routing and control-plane design, performance and benchmarking, model rollouts, and failure handling.
Before working on AI infrastructure, Andrey spent many years building backend and platform systems across SaaS, fintech, payments, data privacy, and high-traffic consumer products.
He writes about AI infrastructure, ML systems, and distributed systems at akrisanov.com.