HAMILTON
Declarative Python dataflows with automatic lineage and DAG execution
ABOUT
Data transformation and feature engineering pipelines often become tangled, undocumented, and hard to debug as they grow — with no clear way to trace which transformations produced which outputs or how changes affect downstream consumers. Hamilton solves this by letting developers declare data transforms as typed Python functions, which the framework automatically compiles into an executable DAG with built-in lineage tracking, visualization, and parallel execution. Each function is independently testable, the full transformation graph is self-documenting, and integrations with Pandas, Spark, Ray, and DuckDB make it suitable for everything from local exploration to production pipelines.
pip install sf-hamilton