Reproducible fleets
NixOS and flakes across servers, workstations and a cloud endpoint. One commit describes the machine; a switch is the deployment.
Open to senior systems & security roles
Systems & Security Engineer
Go · NixOS fleets · vulnerability data · energy infrastructure
I build infrastructure that can be rebuilt from a single commit: Go services with append-only state, NixOS fleets with their own certificate authority, and data pipelines that turn operational noise into a decision. Most of that work has been in energy — telemetry, valuation and risk across renewable portfolios worth more than $100M. The rest is vulnerability data, where CISA's KEV catalogue is the raw material.
Gutach im Breisgau, Germany · remote (CET, US-East overlap)
NixOS and flakes across servers, workstations and a cloud endpoint. One commit describes the machine; a switch is the deployment.
Server-rendered binaries and append-only event logs. No frontend build, no framework, small enough to audit in an afternoon.
KEV as a time series and as work clusters: what is exploited now, and which patch order removes the most risk per hour spent.
One flake for a fleet of NixOS hosts: NAS, workstations, a cloud endpoint and a remote builder, an SSH certificate authority with rotation, encrypted secrets, and a taskfile as the only interface. Every host is commissioned by registering one line.
A single Go binary that turns a request into an immutable, content-addressed specification: append-only events, keyboard-first terminal UI, curl-friendly endpoints, zero frontend dependencies.
A Fibonacci-radix streaming state machine for tokenization experiments: bounded per-step work over a bitstream, adjacency width as a locality selector, Zobrist seed tables, and a corpus benchmark that falsified the seed-only detector.
Quantitative steering of a solar portfolio above $100M: yield and valuation models that survived bank audits, mobile field data capture, and counterparty risk that institutional lenders accepted.
Two public tools make CISA's Known Exploited Vulnerabilities catalogue legible. kev-time plots exploitation over time, so a trend is visible before a report exists. kelly-svg groups entries into clusters a team can work through, so the patch plan follows from data instead of instinct. Both are static SVG and plain HTML, built to be read on a phone during an incident.
I am looking for a senior or lead role where infrastructure and security meet: platform engineering, site reliability, or vulnerability operations. Remote with a CET base, or Freiburg and the Upper Rhine on site.