Benji Lunday

My research:

• (Ongoing) Searching for Stealth Supersymmetry in squark/gluino production channels at the LHC
• Development of an ML-based approach to the Jet Energy Scale (JES) and pileup calibration in ATLAS
• Characterizing static/time-based pixel readout systems for the DUNE far detector

My expertise is:

- Detector hardware characterization and development

• Data processing, analysis, and visualization using Python (pandas, skl, seaborn)
• Developing machine learning networks (DNNs mainly) to tackle physics optimization problems
• Spending hours trying to tweak plots in ROOT

A problem I’m grappling with:

• Simulating events with 6+ jets and multiple interaction vertices in MADGRAPH
• Automating systematics plots for large collections of xAODs/ROOT NTuples/other analysis objects

I’ve got my eyes on:

• Advances in interpretable machine learning and cracking the “black box” problem
• Improving architectures/interfaces for existing physics tools (looking at you, ROOT)

I want to know more about:

Parallel computing, using Numba/JAX in ML contexts, developing more advanced neural nets (especially GNNs), best practices in scripting/automation for large datasets