Before cloud and AI, there was credit risk. Three roles across a decade — an in-house bank, a Big Four consultancy, and a founder-led advisory — building and validating the models regulators require banks to run: PD, LGD, EAD, ECL, IRB, ICAAP and IRRBB.
Preset, configuration-driven test data is fed through a Python statistical validation pipeline that checks model outputs (PD, LGD, ECL) against expected ranges and thresholds, then auto-generates the audit-ready validation report — turning a manual, spreadsheet-heavy exercise into a repeatable, auditable process used across multiple bank engagements.