~/sergey-gamov $ whoami
20 years of operations: from bank mainframes through cloud SaaS to AI-assisted engineering
I am an infrastructure and platform engineer with 20 years in operations. I started in 2006 administering a 200+ server estate at a large regulated bank, moved to a Swiss software company where I built the automation behind its hosted product, spent two years on Kubernetes and CI/CD for an IoT SaaS and a year writing production Python APIs, and came back to infrastructure, now covering platform, network and security for a security-focused SaaS.
I work in small teams where one person owns the whole stack: compute, networks, identity, monitoring, delivery pipelines and the Python tooling that glues it together. I like boring technology, idempotent automation and systems that fail loudly instead of silently.
Based in Germany.
Automate anything you do twice. A manual step is debt that eventually comes back as an incident. The lifecycle tooling I wrote replaced a runbook of manual steps with idempotent, tested code.
Visibility first, optimisation second. Before tuning or re-architecting anything, make sure the basics are measured: hardware, hosts, services. Without that, every migration is a lottery.
Translate requirements into problems. Implementing a feature exactly as requested, without understanding the problem behind it, has cost me hours. Asking "what breaks today?" first has not.
I run Linux infrastructure end to end: virtualisation and bare metal, networks and firewalls, identity and access, backups, monitoring and the delivery pipelines that ship code to it. Configuration lives in code (Ansible, Terraform) and is reviewed and tested like any other code.
When the right tool does not exist, I write it in Python: Django or FastAPI services, command-line tools, integrations with vendor APIs. I keep the tooling small, tested and documented so that the next person on call can read it.
Security is part of the daily work rather than a separate department: encryption, key management, access control, network segmentation and the occasional threat model.
A few things I am proud of, newest first.
Led the migration of a hosted product from server-side encryption to end-to-end encryption on every customer instance: staged rollout, verification after each step and a rollback plan that was never needed.
Redesigned the network of two data centres so that each runs on its own and losing the link between them degrades nothing, plus a segmented office network with an isolated guest segment.
Designed and wrote, as the sole developer, the tool that creates, updates, backs up and retires customer instances. A multi-step manual runbook became one audited, tested command, and no backup restore has failed since.
Production on AWS EKS plus self-hosted clusters for on-demand dev environments, Terraform for the cloud estate, and Docker-based builds that cut application build time by about 30%.
Built a two-stage provisioning pipeline: the customer enters sizing and their own passwords, and ten minutes later receives an encrypted, configured server. The hard part was making sure the initial secret never touched a log or an intermediate system.
Ten years across branch automation, the security department and the core banking systems team: 200+ servers on Windows, Linux and AIX, access control and auditing for regulated systems, an internal Linux mirror adopted bank-wide, and a monitoring rollout that caught problems before users did.
Next I want to go deeper into platform engineering: paved roads for small teams, Kubernetes where it pays off, and security that developers do not have to think about.
AI is changing how this work gets done, and I would rather shape that than watch it. I use AI agents daily for code, runbooks and incident write-ups, and the interesting problems are now on the platform side: giving agents safe, auditable access to infrastructure, deciding what they may touch without a human in the loop, and keeping the blast radius small when they are wrong.
Programming: Python, Bash, SQL, Ansible and Terraform as code, a little PHP and C#.
Human: Russian (native), English (working proficiency), German (learning).