Architecting AI: Efficient Models, Everywhere

Sponsored Session: Architecting AI: Efficient Models, Everywhere - Sharbani Roy, Arm AI is evolving beyond massive frontier models toward smaller, more efficient architectures that deliver real-world impact. In this talk, we’ll explore how Arm is enabling developers to harness Small Language Models (SLMs) and vision models, optimize inference at scale, and build sustainable, responsible AI systems. Through concrete proof points, developer tools, and applied examples, we’ll show how the Arm ecosystem is helping PyTorch developers move faster, reduce costs, and unlock new possibilities across cloud, edge, and device.