Pratik’s research focused on deploying cutting-edge AI and computing solutions to solve large-scale data challenges at CERN’s Large Hadron Collider. The projects demonstrate: deep autoencoders for specialized data compression; optimizing CPU-GPU heterogeneous architectures for a complex real-time processing task; applying pointwise mutual information profiles for anomaly detection in sensor time-series data; implementing zero-knowledge proofs and probabilistic hashing to accelerate and verify AI inference for latency-critical systems; using contrastive VAEs for unsupervised anomaly detection; and building a fine-tuning prototype tool for a RAG system on CERN’s knowledge corpus for enhanced search. This work showcases the application and adaptation of advanced computing paradigms to overcome unique constraints of data volume, latency, and verifiability in a massive scientific infrastructure.
ESR12: Accelerated anomaly detection
Thesis
Planned Title: Advancing computational frontiers for real time analysis in particle physics
Date: April 2026
Publications
Main supervisor: Caterina Doglioni
Caterina Doglioni is a professor of particle physics at the University of Manchester (UK), with a joint appointment at Lund University (Sweden). Throughout her career, she has been driven by finding out more about the constituents of matter as well as by the challenges related to the “big science” needed to study them. The Large Hadron Collider is the perfect scientific environment to combine the two: with her group and colleagues she works on the challenges that a data-rich research environment presents for discoveries of rare processes at ATLAS (more information about dark matter at ATLAS). She has been funded by the European Research Council, first through the DARKJETS ERC Starting Grant, and currently through the REALDARK ERC consolidator grant. She will co-supervise this ESR project together with trigger and tracking expert Jiri Masik at the University of Manchester.
Collaborator: Maurizio Pierini
Maurizio Pierini is a particle physicist working on the CMS experiment at the Large Hadron Collider at CERN. He currently holds an ERC Consolidator Grant for a project that aims to use Deep Learning solutions to address particle physics problems.
LATEST NEWS
CONTACT US
If you wish to contact us,
please use the details below
Email: smarthep-recruitment@cern.ch
SMARTHEP is funded by the European Union’s Horizon 2020 research and innovation programme, call H2020-MSCA-ITN-2020, under Grant Agreement n. 956086
