[Industry Webinar] Edge AI with Executorch on Arm by Matt Cossins
Wednesday 30 September 2026 at 1:00 PM CET, 7:00 AM ET
By Matt Cossins (Developer Ecosystem Manager, AI & Developer Platforms, Arm)
Register now using the link below:
https://attendee.gotowebinar.com/register/2680427286627707481
This webinar introduces developers to the efficient deployment of machine learning models on Arm-based edge devices using ExecuTorch, the lightweight runtime for PyTorch. Designed for audiences ranging from machine learning practitioners to embedded systems engineers, the session explains how ExecuTorch enables AI inference on resource-constrained devices while maintaining high performance and low power consumption.
The presentation begins with an overview of the Arm Edge AI ecosystem, outlining the roles of CPUs, NPUs and GPUs in modern edge AI deployments and positioning ExecuTorch within the broader PyTorch workflow. It then explores the process of converting and deploying models, highlighting key optimisation techniques, including the use of Arm KleidiAI libraries and hardware-specific backends to maximise inference efficiency. A practical demonstration shows how transformer-based models can be executed on a Raspberry Pi using Arm Cortex-A processors, illustrating performance improvements through software optimisation and delegated execution. The webinar also introduces tools such as Model Explorer for analysing model execution and understanding backend delegation.
Finally, participants are directed to hands-on labs, learning paths and supporting resources to continue developing, profiling and optimising AI applications on Arm platforms. The session provides a practical foundation for deploying scalable, efficient edge AI solutions using the PyTorch and ExecuTorch ecosystem.
Presenter’s bio:
Matt Cossins is a Developer Ecosystem Manager in Arm’s AI & Developer Platforms group, where he drives adoption of Arm-powered AI technologies through developer education and technical enablement. His work includes creating hands-on learning resources and delivering workshops for AI and edge deployment workflows, including ExecuTorch. He holds an MEng in Electrical and Electronic Engineering from the University of Nottingham, specialising in AI and neuromorphic computing.