Machine Learning (ML) techniques have been widely used both in science and industry for discovery purposes and for predictions from data. The applications range from High Energy Physics experiments or biological studies to speech recognition or natural language processing and text analytics. Recently, ML methods are gaining interest in the particle accelerator and beam instrumentation community. Task 10.6 has explored the potential of ML-based systems for identifying early signs of errant beam conditions in real-time. This document is the final deliverable of the task and summarizes the acquisition and analysis of ESS commissioning data as well as the development of data workflows and a low latency platform suitable for the training and implementation of real-time ML algorithms at ESS and other high-power accelerator facilities.