Machine learning (ML) can be frustrating and time-consuming since small design decisions can already make the difference between state-of-the-art performance and no learning progress at all. If one does not have ample experience developing new ML applications, it can take days, weeks or even months to figure out the correct settings for all bells and whistles. Automated Machine Learning (AutoML) supports developers and researchers in efficiently obtaining well-performing predictive models. It helps them determine the best hyperparameters, neural architectures, preprocessing, and even entire data processing pipelines for a dataset and metric at hand. This course will summarize the main principles, ideas and recent progress of AutoML, enabling the attendees to understand how to use AutoML efficiently for their next projects. This includes introductions into Hyperparameter Optimization with Bayesian Optimization, Neural Architecture Search, AutoML systems, and human-centered AutoML.