Research

My research sits at the intersection of precision top quark physics, quantum information theory, and machine learning — all within the ATLAS experiment at the Large Hadron Collider.

Quantum entanglement at colliders

The top quark is unique among quarks: it decays before hadronising, so its spin information is directly accessible through the angular distributions of its decay products. This makes \(t\bar{t}\) production a natural laboratory for probing quantum correlations at the highest energies.

With colleagues from Glasgow and Göttingen, I led the first observation of quantum entanglement in top quark pairs, published in Nature in 2024. We measured the entanglement marker \(D\) near the \(t\bar{t}\) production threshold, confirming entanglement at more than five standard deviations. This is the highest-energy observation of entanglement to date.

Current work extends these ideas to the Higgs sector (\(H^* \to ZZ\)) and to tests of Bell inequalities at the LHC — the quantumTANGO project.

Top quark electroweak couplings

Rare processes like \(t\bar{t}Z\), \(tZq\), \(tWZ\) and \(t\bar{t}\gamma\) directly probe the electroweak couplings of the top quark. I have led several ATLAS measurements of these processes, most recently the legacy Run 2 \(t\bar{t}Z\) analysis which includes the first measurement of spin correlations in \(t\bar{t}Z\).

These measurements are interpreted in the Standard Model Effective Field Theory (SMEFT), setting limits on dimension-6 operators involving the top quark. I also prepared the roadmap for future ATLAS+CMS combinations of top+X results in the EFT framework.

Machine learning & anomaly detection

As part of the Dark Machines initiative, I contributed to a large-scale benchmark of anomaly detection algorithms for LHC collisions — testing hundreds of models (auto-encoders, normalising flows, deep sets) on simulated datasets. The goal: model-independent searches for new physics in LHC Run 3 data.

I have also supervised students working on graph neural networks, ML-based lepton isolation, and semi-supervised approaches to BSM classification.

Analysis software for Run 3

At CERN, I am involved in developing the next generation of analysis software for the ATLAS top quark group, targeting the High-Luminosity LHC. This includes the TopCPToolkit — a modular framework for producing analysis-ready data formats from ATLAS PHYSLITE.

Embedded videos

You can embed any YouTube or Vimeo video directly in Markdown pages. Replace the iframe src above with the URL of your actual talk recordings or explainer videos.