Multithreaded (MT) programming is crucial to make best use of today’s parallel computing architectures, but until recently most HEP code was unable to run MT. Because of associated overheads and latency requirements, MT is particularly challenging for RTA. The first objective of the student working on this project will be to implement additional monitoring within the ATLAS real-time code to analyse algorithm scheduling and performance as well as measure the overhead of MT in RTA. The student will use this to identify improvements that maximize the control flow and overall performance. This will be done in synergy with the benchmarking work of another project within SMARTHEP, and integrated in the ATLAS real-time software. Working on this objective will result in the PhD student working on this project becoming trained in advanced techniques of developing and evaluating code for highly parallel architectures. This will be crucial for their collaboration with Lightbox, where they will devise an optimal parallelization of algorithms for investment strategies, trading infrastructures and integrated business processes, and where they will be trained in commercial tasks and then produce a commercial framework with figures of merit for their real-time optimization. The student will also collaborate with CNRS, whose physical proximity enables ESR4 to benefit from the expertise in MT and parallelization more generally of both. The student in this project will use the gained insights and knowledge to implement new RTA capabilities in the ATLAS trigger for Long Lived Particle (LLP) signatures, including dedicated pattern recognition algorithms for charged particles decaying in the middle of the detector. This will increase the trigger acceptance for such particles in ATLAS Run 3 data, so that a search with the first data can be performed in collaboration with another student at the University of Geneva within the SMARTHEP network.