To implement machine learning pipelines, data practitioners must perform several repetitive tasks, for example, to replicate and/or transform data. Aorist is a code-generation tool that aims to automate these tasks. After specifying a universe containing the pertinent datasets, endpoints, user groups, and compliance annotations, the developer can generate a DAG (as a Jupyter notebook or Airflow DAG) containing the Python tasks to manage the data. Under the hood, Aorist will render the code from an underlying AST leveraging knowledge of the target language. In addition, it will compose steps and (predefined) snippets together in an optimal way by using a constraint resolver.