Running AI at the Edge: Systems Lessons from Models to Distributed Intelligence

Abstract: In this talk, I will discuss the evolution of Edge AI from running individual models on resource-constrained devices towards increasingly distributed and adaptive AI systems. I will present some of our recent work on generative AI at the edge, including the deployment and characterization of language and multimodal models, and discuss how model capabilities, resource constraints, and system conditions influence where and how inference should be performed. I will then introduce our work on distributed AI service provisioning and runtime adaptation across heterogeneous edge environments, and discuss some of the systems challenges that arise as AI workloads become more complex, distributed, and dynamic.

Bio: Roberto Morabito is an Associate Professor at DTU Compute, Technical University of Denmark. His research focuses on embedded and edge AI, distributed systems, and resource-aware computing. Previously, he was an Assistant Professor at EURECOM and held research positions at the University of Helsinki, Ericsson Research, and Princeton University.

Read more about the Guest Speaker here: orbit.dtu.dk/en/persons/roberto-morabito/