Learn how to develop and deploy a real-time machine learning model to prevent fake account sign-ups in this informative talk from the Toronto Machine Learning Series. Explore the process of creating an application fraud model using passive biometrics technology and custom-built tools. Discover statistical and computational challenges faced when scaling a low-latency ML inference service. Gain insights into common model deployment pitfalls and learn techniques to address them, including robust model monitoring metrics and the use of a ML feature store. Understand how AWS Glue and Apache Spark can be utilized in building an effective ML Feature Store. Get an overview of the various components required for successful real-time ML model deployments in fraud prevention.