Specialising in Representation Engineering (RepE).
Current research focuses on probing Large Language Models to isolate moral domains, enabling the steering of model outputs toward defined, ethical directions.