Andreas Wagner
System Administrator & Site Reliability Engineer
I make operations predictable for teams without their own ops staff: automated servers, restores that have actually been tested, and monitoring that reports problems before customers do.
- 13 years running production infrastructure, since 2013
- ≤ 1 h maximum data loss (RPO) with hourly encrypted off-site backups
- up to 50 terminals and servers per museum exhibition, managed with Ansible
Profile
System administrator with 13 years of production operations: a PCI DSS regulated fintech, museum exhibitions and my own hosting services. I automate what can be automated (Ansible, runbooks, CI/CD) and document the rest, so that operations don't depend on one person's memory. For me a backup only counts once a restore from it has worked, and monitoring should raise an alarm before a customer does. Because I also develop in Ruby on Rails, I speak the development team's language. I work remotely from Tallinn and come on site when the work calls for it.
Project experience
Container hosting platform for small companies (WellHost)
2022 - todayWhyservices OÜ, own service · Tallinn, remote
Industry: hosting for SMEs and agencies · Role: architect & sole operator · Team: solo
Small companies ran their applications without operations staff: hand-built servers, manual deploys, no separate staging, and backups nobody had ever restored. A failed server or a bad deploy meant days of downtime and possibly lost data. I built a standardised container platform on Docker Swarm, with CapRover as management layer and servers set up by Ansible. Each client gets staging and production separated at infrastructure level, with Backing Services running outside the application containers. Deploys run from git through CI (Woodpecker, GitHub Actions, GitLab CI) with rollback. Hourly encrypted off-site backups are restored in regular drills, and TLS, monitoring and weekly updates are part of the service. At most one hour of data can be lost, and restores are proven rather than assumed. 10 applications (i.e. Rails, Symfony, Django, Wiki.js, GitLab, Plausible) run at 99.95% availability. Clients pay from €50 a month instead of a multiple for dedicated OPS personal.