System Engineer at Hostinger. By day I keep public clouds and Kubernetes clusters from doing what they would naturally do. By night I write the software I wish existed. In between, I'm a dad of one son: the only on-call rotation with no escalation path.
- Dicer · dicer.sh
Dices bare metal into thousands of micro VMs, booted straight from container images. Because containers are great right up until someone says "isolation" in a meeting. - Rungar · rungar.sh
GitHub Actions runners, a fresh machine for every job, on Dicer, Proxmox, GCP or AWS. Your CI deserves better than a shared runner with someone else'snode_moduleson it.
They fit together: Rungar runs each job in its own Dicer micro VM. Call it vertical integration, or call it not trusting anyone else's software.
- ansible-collection-general
Ansible roles for all of the above, because I install things by hand exactly once. - systemd-resolved-exporter
Prometheus metrics for systemd-resolved, because "it's always DNS" deserves a dashboard.
Shorter to list what I haven't worked with:
- a YAML file that was right the first time
- a "temporary" fix that stayed temporary
- Windows Server, on purpose
- a cloud bill nobody asked questions about
- COBOL
- Oracle, because I don't want to be that guy
Everything else, from the kernel up to the Kubernetes control plane, I have probably broken and fixed at least once.
LinkedIn. I reply faster than your
terraform plan.



