Research Interests

I am working on building robust machine learning systems to process the vast deluge of data (exabyte scale) we are about to receive from next-generation telescopes such as the Square Kilometer Array.

Since this much data cannot be curated by humans, we need AI that can deal with messy, real-world data with few labels, unknown data sub-manifolds and out-of-distribution data.

My research has included semi-supervised learning and use of unlabelled data for improved classification in the astronomy domain, developing bespoke contrastive learning methods to apply to astronomical data-sets and use of generative algorithms to simulate astronomical objects.

I enjoy sharing my work with both computer scientists and astronomers and have published my work/given talks at NeurIPS, ICML, EAS and more. A comprehensive list of my publications can be found on Google Scholar.

Broader Interests

I believe that a diverse range of perspectives is key to making breakthroughs and precipitating key insights into difficult technical problems. AI is no exception to this, and as we’ve seen with groundbreaking work such as Alpha Fold, computer scientists have much to learn from scientists and domain experts and vice versa.

I think that we can learn from this success by nurturing further collaboration between the academic machine learning community and those wanting to use AI to solve a domain-specific problem. I always enjoy hearing from people with other perspectives and am open to collaborating on AI based projects.