Hi, I’m Andrey

I'm a backend and platform engineer who likes turning complex systems into something boring, reliable, and easy to reason about.

For the last 15+ years I've been building and scaling SaaS products — mostly on the backend side, often close to infrastructure, and frequently in roles where someone has to take responsibility when things get messy.

Right now, I work on AI platforms and LLM serving: running models on Kubernetes, designing inference infrastructure, setting up observability for GPU workloads, and making sure systems behave predictably in production — not just in demos.

What I care about

I'm especially interested in the space where backend engineering, infrastructure, and applied AI meet:

• LLM serving and inference systems (latency, throughput, cost)
• Production-grade AI platforms, not toy setups
• Observability, evaluation, and failure modes
• Distributed systems that are simple enough to maintain
• Tooling that helps teams move fast without breaking everything

I enjoy work that's close to real users and real constraints — whether that's internal platforms or customer-facing products.

A bit of background

Over the years I've worked in startups and large companies across Russia and Germany, including fintech, data privacy, media, and B2B SaaS.

I've been a CTO, a principal engineer, and an individual contributor — and I'm most comfortable in roles where I can combine technical depth with systems thinking and pragmatic decision‑making.

Somewhere along the way I learned that clean abstractions matter, but boring infrastructure matters more.

Outside of work

I like spending time with my family, reading books and articles on the Internet, and exploring new places and cultures. Nothing fancy, I guess.

This site is a place where I share notes, ideas, and things I've learned — mostly for myself, but hopefully useful to others too.