Making AI deployable in regulated financial institutions — auditable, explainable, and built to pass audit.
I help banks and fintechs put AI to work in their highest-stakes, most governance-constrained workflows: KYC/AML, transaction monitoring, fraud, and identity risk — without breaking what an examiner expects to see.
Eleven years in financial-services risk, controls, and client-facing operations, focused on a single problem: turning AI from a compliance liability into a defensible, reconstructable control.
Applying AI to alert triage and case work while preserving the SAR audit trail and false-positive economics.
Where AI helps and where it fails on synthetic-identity and first-party fraud — and how to keep decisions explainable.
Mapping LLM and agent deployment onto model-risk frameworks: validation, monitoring, oversight, change control.
Designing AI-assisted workflows so every decision is logged, reviewable, and reconstructable under examination.
Every AI recommendation has to answer three questions before it ships: is it auditable, is it explainable, and is it controllable? In regulated finance, a model that can't answer those isn't an asset — it's exposure.
Making AI Pass Audit: A Model-Risk-Management Framework for LLMs in Financial Crime and Identity
How to extend SR 11-7 model risk management to generative AI so it survives examination — auditable, explainable, and controllable.