Notes on dataarchitecture.

Short essays on architecture, cost, hiring and AI: one cartoon, one idea, most days of the week. No tutorials, no listicles.

Orchestration Consolidation Matrix

Count the places a job can be scheduled in your data stack. Include the cron on that one VM.

Count the places a job can be scheduled in your data stack. Include the cron on that one VM.

LIMIT 10 doesn't reduce what BigQuery scans on an unclustered table. A lot of teams learn that from the invoice.

The description fields in your dbt project used to be read by nobody. Now an agent reads them.

The provenance record behind an AI feature fits in a YAML file next to the pipeline code.

Before you replace the data engineer who seems slow, count how many people can hand them work.

Before any stack change I ask how many consumers it touches and whether we can undo it by Friday.

The fastest tools in your data stack are written in a language nobody on your team writes.

GDPR and the AI Act ask their questions at the same moment: when a dataset gets reused for a model.

Novel data projects tend to die between the team that built the prototype and the team stuck running it.

Before your AI feature ships to EU customers, try deleting one customer from everything it reads.

No pitch. An honest conversation about whether I can help, and what shape it would take if I can.