HPC Networking design: Capability to design complex ultra-low latency and high bandwidth networking for HPC and AI architectures, including hardware and the related management software, in collaboration with a leader and other experts in the different areas. * General Networking architectures: Deep knowledge of general-purpose networking architectures, required for the management of storage networks of a HPC or AI cluster. * Networking Ecosystem Management: Ability to interact effectively with leading networking vendors vendors (e.g. Nvidia/Mellanox, HPE/Juniper, Cisco, Arista) * HPC Software: Understand the overall HPC and Ia environemts, and the possible impact of MPI, job schedulers (e.g. Slurm), containerization, AI frameworks (TensorFlow, PyTorch) and other systems management tools (such as Open Nebula or similar ones) when designing the networking infrastructure,
mehr