Notes on evaluating AI systems that actually work — agent evals, data quality, and the architecture around them.https://mayankalok.dev/https://mayankalok.dev/blog/roc-auc-hides-precision/https://mayankalok.dev/blog/roc-auc-hides-precision/Why ROC AUC flatters rare-event detectors, and why the alert budget ended up mattering more than the model.Fri, 24 Jul 2026 00:00:00 GMT