Monitoring Platform Operations
~20M logs per day
Took over a poorly documented production platform on Elastic Stack and Icinga 2 for a nationwide toll operator, then automated it with Ansible and GitLab CI/CD.
Toll operator · via ORDIX
A selection of things I have built and maintain.
~20M logs per day
Took over a poorly documented production platform on Elastic Stack and Icinga 2 for a nationwide toll operator, then automated it with Ansible and GitLab CI/CD.
Toll operator · via ORDIX
Failover: 5 s to 68 s
Compared five PostgreSQL operators for Kubernetes and measured failover under write load. Published as an article (in German) and presented at CloudNativeNight Mainz.
Merge to live in ~3 min
Single-node Talos Linux cluster run entirely from a public Git repository: Argo CD, Cilium, OpenBao, Kyverno and CloudNativePG with point-in-time recovery. This site runs on it.
Replacing CockroachDB
Highly available PostgreSQL clusters with CloudNativePG and Argo CD for the IT service provider of a nationwide retailer, delivered with an operations handbook.
Retail IT services · via ORDIX
Rebuilt weekly, hundreds of participants a year
Terraform modules and Ansible roles that provision training VMs and databases on Oracle Cloud Infrastructure, replacing manual setup by internal IT.
Internal · infrastructure as code
Usability score 32.5 to 80
Bachelor thesis, graded 1.0 (highest possible grade): a domain-specific language that makes modelling the German tax and transfer system in GETTSIM easier and less error-prone.
Static, no client-side framework
Bilingual site with a hand-written WebGL shader hero. Built with Astro, shipped as a small nginx image and deployed to my homelab via GitOps.
Personal project
One API for pandas, seaborn and scikit-learn
University project group: a beginner-friendly Python library, so programming novices can run real Data Science and Machine Learning workflows.