Here are the three main tracks I’m working on right now. Each one is a blend of physics goals, ML methods, and practical software work to make pipelines reliable and reproducible.
Exploring end-to-end approaches that connect generation → detector response → reconstruction, with the goal of validating baselines and identifying research directions (e.g., unfolding-style studies, physics-informed constraints, fast simulation ideas).
Modeling trigger turn-on behavior with neural networks to improve efficiency estimates and systematic understanding. Focused on robust training, careful validation, and uncertainty-minded workflows.
Using ML methods to help detect anomalies in DQM histograms and support operations. Working with matrix factorization / anomaly detection ideas and validating behavior on real monitoring outputs.