My research focuses on developing ML approaches tailored to optimising particle track reconstruction algorithms at
the Large Hadron Collider experiment at CERN. Working in a diverse and collaborative environment gave me the opportunity to learn from experts in the field
and allowed me to develop a foundation of interdisciplinary scientific knowledge, combining intuition with creativity!
Leveraging GNN Architectures
The approach aims to iteratively extract particle track candidates by utilising Graph nodes, Gaussian Mixture Reduction techniques and embedded Kalman filters within network message passing.
Pre-print article available here.
Working at CERN
During 2022-2023 I worked at the LHC experiment in Switzerland, where I was a part of the ML Physics community.
The ATLAS detector,
fundamentally a complex Big Data project, is designed to observe up to 1.7 billion particle collisions per second, with over 100 million
electronic readout channels and a combined data volume of more than 60 million MB/s.
ML Predictor for Measurement-to-Track Association
Design and construction of a supervised ML classifier used to identify compatible hit-pairs of particle trajectories using geometrical input features.
The method employs a Naive Bayes approach, with fitted kernel densities for the likelihood function and uses a fast lookup implementation in order to balance trade-off concerns
between efficiency metrics and computing resources. This approach was deployed into production software, achieving over 2x speed-up factor.
Published article available here (2023).
