At sufficient scale everything fails all the time, and the only real feature is being able to watch it happen.

_{— me, to a 🦆}

I work on the metric backend of Checkmk, its data backend for OpenTelemetry metrics that lets it monitor applications, not just infrastructure.
Before that I was at SAP, building the CI/CD infrastructure for HANA (their in-memory database): a graph-based task execution framework running on ~2000 compute nodes spread across multiple cloud providers and regions.

since 2021            csshW: broadcast your keystrokes to all your SSH sessions (successor: cssh-rs)
PyConDE '25            Rustifying Python: what's worth moving to Rust, and what will go wrong

2026-09-08 17:25 CEST  Checkmk/checkmk log: render ui-job-scheduler server logs in the new…
(today)                Checkmk/checkmk _{507 commits}

2026-09-08 17:09 CEST  Checkmk/checkmk cmk-astrein: add component ownership
(today)                Checkmk/checkmk _{507 commits}

2026-09-03 22:49 CEST  whmade/cssh-rs  client: ConPTY DSR responder always answers cursor…
(thursday)             whmade/cssh-rs  _{566 commits · 118 pull requests · 152 issues}

2026-09-03 22:45 CEST  whmade/cssh-rs  client: size the ConPTY to the console viewport and…
(thursday)             whmade/cssh-rs  _{566 commits · 118 pull requests · 152 issues}

_{Last 30 days}

Lua 47.5% (9.6k) · Rust 14.9% (3k) · Python 10.6% (2.1k) · TypeScript 10.0% (2k) · PowerShell 8.8% (1.8k) · Vue 6.5% (1.3k) · Other 1.8% (360)

_{All time}

JavaScript 24.8% (71.3k) · Python 24.6% (70.8k) · TypeScript 14.3% (41k) · Rust 13.3% (38.3k) · Vue 10.8% (31.1k) · Lua 5.0% (14.4k) · Other 7.1% (20.3k)

_{Last updated 2026-09-08 19:15 CEST.}

_{Everything above refreshes every few hours via a small workflow. Here's how it's computed.
The 🦆 is maintained by hand.}