Arm at the PyTorch Conference 2025

Arm is attending the PyTorch Conference on October 22-23 to demonstrate how its ecosystem enables the development and deployment of AI applications using PyTorch and ExecuTorch. The company is focusing on providing developers with tools and feedback loops to optimize AI workloads across cloud, edge, and mobile environments.

AI Deployment and Optimization Tools

Arm is showcasing a variety of optimized use cases and technologies at Booth P1, focusing on the transition of AI workloads from the cloud to the edge. Key technologies and frameworks being demonstrated include:

  • ExecuTorch: Used for deploying AI on mobile, gaming, and Edge AI platforms.
  • vLLM and Mixture of Experts (MoE): Demonstrated for cloud-based AI deployments.
  • Interactive Demos: Showcasing neural graphics training, audio generation, speech recognition workloads, and agentic AI workflows.

Developer Support and Product Refinement

Arm is offering specialized one-on-one workshops and feedback sessions to help developers improve the usability and safety of their AI products.

Usability and Responsible AI Workshops

Design experts are providing personalized sessions to refine user interfaces, streamline user experiences, and implement intuitive workflows. A central component of these workshops is "Yellow Teaming," a proactive approach used to identify and mitigate risks related to fairness, transparency, privacy, and security before AI systems are deployed.

Voice of the Developer Sessions

Arm is conducting 30-minute one-on-one conversations to gather data on how engineers build, deploy, and scale AI. These sessions specifically target real-world development challenges, including:

  • Migrating workloads from NVIDIA/x86 architectures to Arm.
  • Profiling and debugging models at scale.
  • Running Large Language Models (LLMs) reliably at the edge.

The insights from these sessions are intended to inform the development of Arm's future AI tools, SDKs, and documentation.

Sources