Kevin Pedro

Kevin Pedro

Associate Scientist

Telephone: (630) 840-5686   E-mail:  Office Location: WH11E



Scientific Interest

I first joined the CMS collaboration in 2009, and have also worked on Daya Bay, ATLAS, and CLIC. In CMS, I was previously the convener for the offline software for the HCAL Phase 1 upgrades. I now coordinate the offline software for all CMS Phase 2 detector upgrades. I am highly involved in the intersection of AI and computing. The SONIC project, for which I am the lead developer, uses coprocessors, such as GPUs and FPGAs, as a service to accelerate neural network inference. I am also investigating the use of graph neural networks for calorimeter clustering and considering how to use AI to speed up detector simulation.

At the start of Run 2 of the LHC, I focused on searching for strongly-produced supersymmetry in events with jets and missing energy. My physics interests now focus on dark matter; specifically, I am searching for evidence of strongly-coupled hidden sectors. These models can produce semi-visible or emerging jets that contain invisible and/or displaced particles. Such signatures have generally been ignored by existing collider searches, providing the potential to discover the nature of dark matter at the LHC. I am exploring AI as a powerful tool to extract these signals from large, noisy backgrounds.