Model risk auditing for fraud detection research and deployment pipelines

This is an AI quality-control workflow that checks whether a fraud model was trained and tested fairly, so a bank does not trust a model that only looks good because of mistakes in the pipeline.