Amazon Machine Learning provides developers with visualization tools and wizards to help create ML models without having to learn complex ML algorithms or manage the underlying infrastructure. It allows you to start small and then scale, as your application grows. Typically, after the modeling phase is completed, Amazon Machine Learning makes it easier to obtain the results using simple APIs. These APIs assist in serving these results in real time. You can also use Amazon Batch for processing predictions based on large batches of data. 

Apache Spark is an open source, distributed processing system that is increasingly becoming the platform of choice for many emerging applications, including streaming analytics, machine learning, and graph applications. You can quickly create managed Apache Spark clusters and use auto scaling to increase or decrease the size of your clusters.