Data & AI
Analyses that reproduce, models that are measured, constraints that hold.
Requires The Book of Works
INotebook to pipelineThe analysis that ran once in a notebook is a rumor. Parametrize it, track it, make next quarter free.IIEvaluate the modelVibes are not a metric. Build the eval, measure the misbehavior, re-run when anything changes.IIITrack every runAn experiment that didn't record its config and result may as well not have run.IVSchema, not promptPrompts handle fuzzy judgment. Hard constraints live in schemas, tools, and filters, where they cannot be ignored.