Hamilton is a general purpose micro-framework for creating dataflows from python functions. It simplifies the creation of wide column dataframes and works on python objects of any type and dataflows of any complexity. The core design of Hamilton maps function names to components of the generated artifact, making it easy to understand the relationship between code and data. It allows for easy modifications, self-documenting code, and unit testing of data transformations. Hamilton functions form a Directed Acyclic Graph (DAG), which can be executed, optimized, and reported on by the Hamilton framework. This repository is maintained by the original creators of Hamilton, who have since founded DAGWorks Inc., a company dedicated to building and maintaining the Hamilton library. For more information on the origins of Hamilton, refer to the original Stitch Fix blog post. Hamilton is compared to other macro orchestration systems like Airflow, Feast, dbt, and Dask. It offers features such as Python 3.7+ support, code structuring, unit testability, documentation friendliness, easy lineage visualization, library-only usage, platform compatibility, python transformation management, and support for GenerativeAI/LLM based workflows. However, it does not replace macro orchestration systems and is not a feature store. To try Hamilton without installation, visit www.tryhamilton.dev. If you need further assistance, join the Hamilton Slack community.
