Rayen Dhahri holds a Master’s degree in Robotics, Cognition, Intelligence from TU Munich. His thesis, supervised by Dr. Vincent Fortuin, Alexander Immer and Bertrand Charpentier, focused on developing methods to enhance the efficiency and compression of neural networks. Currently, Rayen is expanding his thesis work to encompass larger neural networks and modern architectures, aiming to demonstrate the generalization of their proposed method on newer architectures. He worked at BMW, deploying neural network models on microcontrollers, optimizing architectures for real-time applications, and generating custom kernels for ECUs. At Huawei, he developed a reinforcement learning fleet management framework, optimizing task assignments for robots based on available resources and task requirements, while enabling dynamic learning of new tasks. His research projects include proposing a teacher-student architecture for saliency prediction combined with data augmentation techniques at EPFL, developing neural network sparsification methods at Intel, combining hardware filters and machine learning for anomaly detection at Infineon Technologies, and generating kernels using LLVM for deploying models on RISC-V, contributing to the creation of an ELF Loader and optimizing memory layout for an Instruction Set Simulator.