UAT regression framework for billing system upgrades. Tolerance-aware diffing of pre/post-upgrade report extracts: 100% catch rate measured against 1,299 planted regressions across 7 classes, zero false alarms on a benign-only control, 1M rows in 67s, and a formatted Excel sign-off evidence workbook as the first-class artifact.
-
Updated
Jul 16, 2026 - Python