Every user has a unique digital signature with respect to typing, moving the mouse or accessing a web page. For instance, by analyzing the time between specific keystrokes of many users, we can verify whether user A really is user A or whether she/he has assumed the identity of another user. The goal of this project is to develop an advanced analytics foundation based on Qumram’s data. The idea is to create directly sellable products on top of this foundation to identify individual users on the digital channels with high accuracy. In particular, we apply machine learning and anomaly detection algorithms to analyze the recorded online behavior of specific users and perform automated risk class scoring. The main challenge is to perform the algorithms in real time by leveraging state of the art big data technology such as Spark Streaming and Spark Machine Learning.