LPC Distinguished Researcher


Patrick McCormack



What I will be working on:

I will be working on integrating GPUs into CMS workflows to accelerate algorithms via an Inference-as-a-Service scheme called SONIC ("Services for Optimized Network Inference on Coprocessors"). During my time as a DR, we will focus on making SONIC ready for full-scale deployment in default in CMS offline workflows. I will also work on machine learning-based anomaly detection techniques for discovering new physics at the LHC.

My role in CMS past and present:

I joined CMS in 2021, having completed my graduate studies on ATLAS at UC Berkeley. Since joining, I have worked on development and testing of the SONIC scheme and have participated in an anomaly search using CMS run 2 data. In this search, we deployed the Quasi-Anomalous Knowledge (QUAK) algorithm. I have also worked on creating an ML-based HCAL clustering algorithm for CMS using the Sparse Point Voxel CNN architecture.