Advisor: Alessandro Abate

Nandi Schoots is a postdoc at the OXCAV group at the University of Oxford. She focuses on utilizing the science of deep learning for model control before and during training. Her interests are broad and range from mitigating steganography and collusion between agents to identifying functional equivalences between different model representations. Formerly she was part of the Safe and Trusted AI group at King’s College London and Imperial College London.