About Me

Hi! I’m an incoming second-year PhD student at the MIT Operations Research Center. My research lies at the intersection of AI safety, evaluation, and optimisation, with a particular interest in multimodal and generative AI.

My current research interests span three broad directions:

1. Non-generative safety evaluation: developing methods to identify harmful model capabilities without producing harmful content, with applications to child sexual abuse material (CSAM), non-consensual intimate imagery (NCII), and biological misuse.
2. Epistemic harms from AI: studying how hallucinations, sycophancy, homogenisation, and overreliance on AI-generated information affect what people know, believe, imagine, and regard as possible.
3. Efficient and responsible machine learning: reducing the computational and energy demands of modern AI systems without compromising their safety or reliability.

Across these directions, I hope to develop rigorous methods that improve the human condition, positively affect people’s lives, and deepen our understanding of how increasingly powerful AI systems behave and influence the world.