Shavindra Jayasekera

Shavindra Jayasekera is a PhD student in the Modern Statistics and Statistical Machine Learning CDT programme at Imperial College London, which he joined in October 2023. He completed his undergraduate and master's degrees in mathematics at the University of Cambridge, specialising in statistics, and was awarded high distinction for his Part III Essay on Generative Adversarial Networks in Biomedical Imaging. He then received funding from the Long-Term Future Fund to complete a one-year master's in computational statistics and machine learning at University College London. His doctoral research at Imperial focuses on improving the explainability, interpretability, and reliability of foundational models, with additional interests in machine learning safety and applications in chemistry. He was a co-author on the paper "Variational Uncertainty Decomposition for In-Context Learning" presented at NeurIPS 2025, and was selected as one of six finalists for the IMA Lighthill-Thwaites Prize in 2025.

No endorsements yet.

Personal Website

-

LinkedIn

LinkedIn

Twitter / X

-

LessWrong

-

EA Forum

-