Hannah Cyberey

I’m a Postdoctoral Research Associate in the School of Data Science at the University of Virginia, supervised by Alex Gates. My current research develops methods for measuring how AI capabilities emerge and how research translates into open-source AI infrastructure.

My research focuses on the evaluation and auditing of language models: testing whether widely used metrics and benchmarks measure what they are intended to measure, and developing white-box auditing methods that surface risks standard evaluations miss. I received my PhD in Computer Science from the University of Virginia, advised by Prof. David Evans and Prof. Yangfeng Ji. My dissertation introduced sensitivity auditing with steering vectors, with applications to detecting and controlling model behaviors, such as refusal, censorship, and bias.

Email: yc4dx at virginia dot edu