Synthetic Data Helps Close Gap Between Fraud Detection and Legitimate Payments

For Adarsh Naidu, a Senior Solutions Architect in Amazon Web Services' Enterprise Banking group, fraud detection is not just about catching criminals but also about deciding which of millions of normal purchases to let through. A system tuned too tightly to catch every possible fraud will block legitimate spending, while one that waves everything through will miss real threats.

Naidu has spent over two decades working on this problem in financial services firms like American Express and Assurant before joining AWS. He's developed a technique to train fraud systems on synthetic data, which lets them rehearse against a wider range of scenarios without pulling live customer information into every experiment. This approach keeps privacy intact and helps the system tell apart legitimate spending from real problems.

Naidu applied this same method to dispute handling, publishing a reference piece in 2024 for the AWS for Industries blog on dispute management in banking. He's also authored a Responsible AI framework for AWS that builds governance into each stage of system development rather than adding it at the end.