Real-time Machine Learning is a technique of continuously improving a machine learning model by training it with real-time rich data. These models can be deployed to production using event-driven architectures in which rich data streams are fed into these models by combining data-at-rest with data-in-motion. However, are we there yet? Deploying real-time machine learning models has its own challenges: such as complexity, scalability and performance. The talk will address these challenges and will demonstrate best practices for real-time machine learning using the Hazelcast open-source platform, the demo code will be hosted on Github.