IT consulting · DevOps & infrastructure · Berlin
Infrastructure you can stop thinking about.
I design, automate and run infrastructure for companies that would rather build their product than babysit servers: bare metal and cloud, Kubernetes, CI/CD, databases, observability. Fifteen years in operations, from ad-tech platforms that take billions of requests a day to fleets of blockchain nodes. Based in Berlin, working remotely across Europe.
Open to a full-time staff role, and taking on consulting projects. CV →
whoami oleg — Staff DevOps / SRE engineer, Berlin uptime up 15 years, load average: bare metal, Kubernetes, PostgreSQL, ClickHouse cat ~/focus automation · reliability · observability · plain language ls ~/projects ump/ github.com/ofzhur
What I do
Infrastructure work from the rack to the deploy button, for teams of one to fifty. Hands-on when something needs building, advisory when a decision needs a second opinion.
Cloud and bare-metal infrastructure
Hetzner, OVHcloud, Leaseweb and the hyperscalers, or your own hardware: sizing, network and storage layout, provisioning, hardening and the day-two routine. Bare metal where it pays off, cloud where it is quicker, and migrations between the two with a plan, a dry run and a way back.
Kubernetes and containers
Clusters sized for the workload rather than the hype: setup and upgrades, ingress and storage, operators, and the images and charts your services ship in. Or plain Docker on a few machines, when that is the honest answer.
CI/CD and infrastructure as code
Pipelines on GitHub Actions, GitLab CI or Jenkins that build, test, release and deploy without hand steps. Terraform and Ansible for everything underneath, so that every change is reviewed, repeatable and easy to undo.
Databases and data platforms
PostgreSQL and ClickHouse in production, from one instance to petabytes of analytics data: replication, backups that get restored on purpose now and then, upgrades with as little downtime as the setup allows, and the tuning that lets a platform take billions of requests a day. Redis, Kafka, Elasticsearch and the rest of the pipeline around them.
Observability and operations
Prometheus and Grafana, alerts that wake the right person for the right reason, logs you can actually search, and incident reviews that end in a fix rather than a culprit. Runbooks instead of folklore.
Consulting and architecture
A review before a decision, a roadmap for a small team, an audit of what you already run, a look at a bill that grew while nobody watched, or a few days a month as the ops person a product team does not have yet.
Projects
Tools I build when the same problem shows up at every client, and where the rest of my code lives.
UMP — Unus Mundus Provisioner
One operating system, in the same shape, on any server: a CLI that installs the same Ubuntu, with the same disks, users, keys and firewall, on dedicated servers and cloud VMs at Hetzner, OVHcloud and more. Open source, in Go.
ump.zhurko.de →
GitHub
The rest: open-source odds and ends, from SaltStack formulas to Kubernetes operators and small command-line tools.
github.com/ofzhur →
How I work
The same principles whether the job is a two-day review or a year of running your platform.
- Everything is code
Servers, clusters, pipelines and DNS live in a repository, reviewed and versioned. Nothing depends on what one person remembers.
- A plan before the change
Dry runs, staged rollouts and a tested way back. Destructive steps are announced, written down and confirmed, never improvised.
- Boring or bold, on purpose
Proven tools, fewer moving parts and safe defaults for everything that has to run at three in the morning. The newest technology, or something nobody has tried yet, where that is what makes the rocket fly. Either way the choice is deliberate, not a habit.
- A handover you own
Documentation, runbooks and a walkthrough come with the work. You keep the keys and the knowledge; I am glad to stay on, but you never have to keep me.
Toolbox
What I have run in production and reach for first. The list is not the point; the right tool for your problem is.
Platforms
- Linux (Ubuntu, Debian)
- Bare metal
- Hetzner
- OVHcloud
- Leaseweb
- servers.com
- AWS
- Google Cloud
- DigitalOcean
Containers
- Docker
- Kubernetes
- Rancher
- kops
- ECS
- Operators in Go
Automation & CI/CD
- Terraform
- Ansible
- SaltStack
- Chef
- GitHub Actions
- GitLab CI
- Jenkins
- TeamCity
- Drone
Databases & data
- PostgreSQL
- ClickHouse
- Redis
- MongoDB
- MySQL / MariaDB
- Elasticsearch
- Kafka
- RabbitMQ
- Hadoop & Spark
- Druid
Observability
- Prometheus & Thanos
- Alertmanager
- Grafana
- ELK
- Sentry
- Graphite
Network & edge
- nginx
- HAProxy
- Cloudflare
- Route 53
- OpenVPN & strongSwan
Code
- Go
- Python
- Ruby
- Bash
- SQL
Blockchain
- Ethereum nodes (geth)
- BNB Chain nodes
- Multi-chain explorer infrastructure
Clients
Companies whose infrastructure I build and run, with the DevOps around it: servers and clusters, pipelines, monitoring, the day-to-day. On retainer and in projects.
Turbologo
Infrastructure & DevOps
An AI logo maker and brand-identity platform.
turbologo.com →
PostPost
Infrastructure & DevOps
Brand-first AI for social media: on-brand posts, carousels and videos, scheduled and published from one workspace, with the analytics to match.
postpost.ai →
Thryve
Infrastructure & DevOps
Health-tech from Berlin: one API for the data of hundreds of wearables and medical sensors, for insurers, pharma and clinical research.
thryve.health →
Go X
Infrastructure & DevOps
E-scooter sharing in US tourist destinations, with a B2B platform for partners who host fleets or run their own.
goxapp.com →
Let’s talk about your infrastructure
Tell me what you run, what should be better, and by when. I read every message myself and reply in English or German.
Hiring for a staff role? See my CV.