How Wyze Uses AI to Scale Cost-Efficient Video Understanding

Since its founding in 2017, Wyze has become a leading force in the smart home industry, delivering over 30 million AI-powered devices to more than 10 million households across the U.S. Every day, Wyze's AI infrastructure ingests and processes billions of video events - presenting enormous challenges in secure, accurate, and cost-efficient video processing at scale.
In this talk, we will explore how Wyze AI is tackling these challenges across multiple dimensions.

Edge-to-Cloud Video Processing: Strategies for efficiently distributing and processing high-volume camera footage between on-device and cloud systems. 

Deep learning based AI detection: How earlier generations of AI have been successfully deployed to provide powerful person, pet, package, and vehicle detection (PPPVF) features at low cost to customers.

LLM enhanced video understanding: How the latest advancements in large language and vision-language models (LLMs and VLMs) are enabling semantic video search, deeper event understanding, and enhanced customer experiences.

This session will offer a deep dive into how Wyze is building one of the largest and most intelligent smart home video platforms in the U.S., highlighting the practical AI innovations that are shaping the future of home security, automation, and user experience.