Scientific Machine Learning Physicist

Develop and evaluate graph neural networks in the NuGraph framework for the DUNE neutrino experiment. Combine charge-detector and photon-detector measurements to classify low-energy interactions and improve event reconstruction.

Extend PyTorch research code with optical graph inputs and cross-modal detector connections. Prepare simulated training data, create graph samples, manage data splits, train and validate models, and evaluate results with confusion matrices, probability outputs, and event-level error analysis.

Scale classification studies to 40,000 simulated events by processing and combining HDF5 datasets on shared computing facilities. Develop regression decoders for interaction-vertex and electron-direction reconstruction, and study the effects of graph inputs and loss functions.

Lead neutron-source calibration and photon-detection studies with Fermilab and CERN collaborators. Present results at collaboration meetings and conferences, coordinate analysis activities, and mentor graduate and undergraduate researchers.

Led cosmic-muon calibration studies for the ProtoDUNE-SP liquid-argon detector, developing charge and energy calibration methods for physics measurements. Measured pion-argon interaction cross sections for doctoral research using event selection, detector corrections, simulation comparisons, and uncertainty studies.

Investigated space-charge effects and electron diffusion to improve detector-response interpretation. Developed reusable analysis and visualization software in C++, Python, and ROOT, communicated results to the international DUNE collaboration, and taught undergraduate physics and laboratory courses.

Dissertation: A pion-argon cross section measurement in the ProtoDUNE-SP experiment with cosmogenic muon calibration.

Completed training in databases and SQL for data science with Python.