This document discusses SKIL, an intelligence layer from Skymind that provides tools for the deep learning workflow including exploratory data analysis, model training, deployment, monitoring, and scaling. It focuses on enterprise deep learning and provides infrastructure for using, serving, and visualizing deep learning models while auditing data flow. Training models is described as difficult due to the lack of interpretability in neural networks and emphasis on research over applications. Deployment options on DC/OS or Docker aim to provide scalability and a consistent development and production environment.