Keynote

We are witnessing a transformative shifts in computing, giving rise to the 3D computing continuum, an integrated ecosystem spanning Cloud, Edge, and recently orbital computing systems such as LEO satellites. This novel computing landscape enables seamless, concurrent computation and data processing across heterogeneous tiers, optimizing for latency, performance, and resource efficiency. Yet, it also brings new challenges: managing heterogeneous and constrained devices, coping with intermittent connectivity, and ensuring resilience in harsh environments, including exposure to space-specific conditions like solar radiation.

At the same time, Serverless computing has emerged as a compelling computing paradigm for developing next-generation, continuum-native applications. It represents the next step in the evolution of cloud computing and introduces novel programming models (FaaS), execution models, services, and platforms. Serverless computing is especially appealing due to its low infrastructure management overhead, easy deployment, scale-to-zero, and the promise of optimized costs. However, it also introduces its own challenges including cold starts, managing stateful functions and guaranteeing stable performance.

Atop this evolving infrastructure, Compound AI is emerging as the next frontier of artificial intelligence. These systems combine diverse AI/ML components, including custom and off-the-shelf models, vector stores, retrieval pipelines, and coordination algorithms into self-optimizing, adaptive AI networks. Much like the human brain’s specialized regions working in unison, Compound AI integrates modular intelligence to handle sophisticated tasks. However to realize its full potential, we must rethink models, system architectures, and orchestration techniques so that future AI systems can benefit from the emerging computing paradigms and computing landscapes.

In this talk, we explore the confluence of the Compound AI, serverless computing paradigm, and the 3D computing landscape, particularly focusing on the opportunities, challenges, and best practices for building such novel systems.  We specifically examine how the non-functional requirements such as performance, scalability and reliability are addressed at this intersection of next-generation computing paradigms, computing landscapes and distributed AI systems.