In this workshop we will study the application of deep learning to banking analytics, studying the different types of unstructured and structured data common in banking analytics (in risk management, marketing, fraud detection, etc) and the different architectures and models which are available to make use of these sources of data. We will also study the statistical and business-oriented performance measures which can be used to evaluate these models. We will go beyond common single data source models to combine multiple architectures in multimodal (or multichannel) models, and we will study the implications of different combinations of data sources possible in a deep learning model. Bring your laptop or tablet for an interactive workshop using cloud infrastructure, where we will implement a multimodal analytics model from start to finish, using different types of embeddings, text data, and structured data, studying the impact of our architecture decisions.