Facial Detection for Test Security

Overview

High-stakes testing depends on knowing who is actually in the seat. At ETS I coded a facial-detection system in Python to strengthen identity verification in test centers — replacing a heavily manual, expensive security process with an automated one.

Impact

The system sharply reduced security costs and gave ETS the leverage to negotiate more favorable contract terms with its testing centers. It’s a favorite example of applying ML to an operational problem where the business case was immediate and measurable.