Bradley Leesman

This project investigates how efficiently HEP datasets can be used in modern AI and machine-learning workflows. It benchmarks data loading and I/O performance across HEP data representations and examines the overhead introduced by data conversion and preprocessing. The project characterizes data-access behavior at scale and identifies bottlenecks that limit the direct use of HEP datasets for AI. The results will help guide the development of more AI-ready data formats and workflows for large-scale HEP applications.