Writings
Notes from the workshop.
We earn authority by writing what we believe — and being wrong in public when we are. Architecture, SRE, SecOps, and the politics of platform engineering. RSS available .
-
Paved roads vs platforms: what we learned getting it wrong
Why 'narrow opinionated tools' beats 'broad coherent platforms', and the four times we relearned this lesson expensively.
Read -
SLSA L3 without buying a SaaS
A self-hosted attestation pipeline using sigstore, GitLab CI, and zero new vendors. With the YAML.
Read -
Error budgets that don't lie about themselves
Most SLO programs fail by Q3 because they were never tied to a real consequence. The contract template we use.
Read -
Picking the smallest model that does the job
A reproducible eval workflow for choosing between vLLM, SGLang, and llama.cpp under a fixed VRAM budget.
Read -
How to write an ADR no one will read (and how to write one they will)
A short essay on style, structure, and the political surface of decision records.
Read -
Service boundaries through MCP contracts
Why MCP makes a better contract surface than gRPC for agent-shaped systems, and where it breaks.
Read -
Threat models you can fit on one page
STRIDE is a worksheet, not an artefact. We argue for one-page threat models with explicit non-goals.
Read -
On-call rotations that don't burn out the senior engineer
A staffing model that keeps page rate under 2/week without sacrificing escalation paths.
Read -
Why we don't publish customer logos
An argument for authority through public reasoning instead of borrowed brand equity.
Read