In this role, you will contribute to the development of machine learning-based Particle-Flow reconstruction for the CMS experiment, integrating advanced algorithms into the High Level Trigger as part of the Next Generation Trigger project, where timing constraints and real-time performance are critical.
Responsibilities:
Develop ML-based PF components using TICL inputs in CMS and validate performance using standard PF and TICL metrics.
Ensure robustness, interpretability, and debuggability in realistic CMS environments.
Explore the applicability of these approaches for future collider detectors, building on the FCC framework and the Key4hep ecosystem.
Lead ML-based reconstruction studies for CLD, and extend the approach to other detector concepts such as ALLEGRO, IDEA, and GRAiNITA.
Designing suitable data representations for heterogeneous detector inputs.
Handling large-scale graphs and distributed training.
Benchmarking performance on physics observables and reconstruction metrics.
Requirements:
You are a national of a CERN member or associate member state. A limited number of positions are also available to candidates from non-member states.
You have a professional background in computer science, physics, or a related field and have either
a Master's degree with 2 to 6 years of post-graduation professional experience;
or a PhD with no more than 3 years of post-graduation professional experience.
You have never had a CERN fellow or graduate contract before.
Demonstrated proficiency in detector physics, event reconstruction principles, and physics analysis in the context of high energy physics experiments is essential.
Strong experience in advanced ML model creation, large-scale and distributed training, and deployment is required, as the role involves developing and incorporating AI-driven techniques into the reconstruction algorithms.
Strong programming skills in Python are necessary for scripting, tooling, and integration tasks.
Strong programming skills in C++ are required, with a focus on developing efficient algorithms and, eventually, integrating different ML into the HEP framework for fast inference; familiarity with CMSSW and FCCSW is a plus.
