Marcin Pyrka

Engineering manager. I run observability at extreme scale and build governance for MCP — the boring, load-bearing layers everything else quietly depends on.

Monitored endpoints

checked 2026-08-03 12:11 UTC · at build

MCP HangarMCP policy enforcement plane. Open source, self-hosted, no SaaS.operational
whyisthisdown.comWriting on observability, incident forensics, and MCP governance.operational
GitHubCode, issues, and the occasional strongly worded ADR.operational
LinkedInProfessional profile and contact.corporate

Recent write-ups

The Fossil Record2026-07-31
The Bill Came Due Early2026-07-28
There Was No Tenant2026-07-02

Incident history

A decade and a half of large-scale platforms: Hadoop pipelines in pharma, an IIoT PaaS with 15+ engineers, and the observability stack behind one of Europe’s largest e-commerce platforms — VictoriaMetrics at a scale where “cardinality” stops being a word and starts being a lifestyle.

These days the interesting failure domain is agentic infrastructure. I build MCP Hangar, a policy enforcement plane for MCP traffic, and I document what breaks along the way at whyisthisdown.com. Working rule for both: verify on the artifact, not the changelog.