38The model aced every offline eval and the customer says it's useless. The labels look fine. What now?▼hardScaleGoogleDatabricks1 replies◆ premiumNot leakage, not skew, not drift: the model learned exactly what you asked, and you asked for the wrong thing. The proxy-label mismatch that passes every test, the audit that catches it, and why this is the failure no metric can see.Open full answer →
69A bank wants one fraud model across three systems it acquired. Each one labeled fraud differently. Scope the first 90 days.▼hardNewPalantirScaleDatabricks2 replies◆ premiumEveryone spots that the customer records need resolving. The failure that actually sinks this project is quieter: the word fraud means three different things in the three datasets, so the union of their labels trains a model that is confidently wrong.Open full answer →
