Open Source Summit + Embedded Linux Conference Europe 2020

Machine learning models usually make predictions based on data coming from a wide range of IoT devices. If we think of images, audio recordings or brain waves we can see that they are all measured using hardware sensors. After being read, this data is usually sent to remote clusters where inference is performed. Wouldn’t it be great if we could expand these devices to also make predictions? Edge computing can help to address the privacy, latency and data ownership concerns by bringing this computation to the “edge”. In this talk we will discuss these concerns and we will introduce KubeEdge as a solution to treat our edge devices as Kubernetes nodes, which will enable us to use existing Kubernetes tools to deploy machine learning models and perform real-time inference.