I am a Senior Lecturer (~Associate Professor) in Artificial Intelligence in the Department of Computer Science at City St George's, University of London and a Principal Investigator at The Alan Turing Institute.
My group builds language models and agents that reason reliably. We combine neural models with symbolic structure, such as grammars, logic programs and answer-set solvers, so that what a model generates can be controlled and checked. We also study how to evaluate and verify model outputs, and how models reason over multimodal data such as charts and visualisations. A thread running through this work involves cognitive elements, in that we ask: what human constraints on memory and belief can teach us about how models should learn.
Earlier, I worked on multimodal machine translation, grounded language learning, and structured prediction for parsing. More on my research →