This presentation reveals how Ansible can be a critical and powerful tool in making large-scale quantitative research and machine learning teams productive. As an example of this, we will show how Ansible became a major architectural element in the platform design for distributed quantitative research teams. Starting with our goals of “Speed to Prediction” and “Cost per Prediction” we built a scalable platform for developing, testing, and productionizing machine learning models at scale. With Platform our researchers focus primarily on hypothesis development and testing and Platform takes care of everything required to run and scale models with high availability. This presentation explains how Ansible helps our team to run as fast as possible. It will also reveal our lessons learned while using Ansible automation for testing ML models, provisioning ephemeral nodes, as well as running models in production.