Although libraries like Pandas make it easy for data practitioners to create dataframes, managing dataframes (which may specify features on which a model is trained) in a common code base and at scale can be challenging. To facilitate this task and make complex data pipelines less brittle, Stitchfix developed and OSS’ed Hamilton- a microframework for generating dataframes from specifically shaped Python functions. With Hamilton, the name of the function is the name of the column and the input parameters are the name of the input variables. As such, Hamilton can use the function definition to create and execute a DAG (which represents the relationships between columns). With Hamilton, documenting and testing dataframes is also much easier.