A declarative language for data pipelines.

BaruchML is a markup-language-inspired method for connecting, transforming, and outputting data. Describe the pipeline; let the runtime work out how to run it.

Read the docs View on GitHub

What it looks like

source:
  orders:
    type: csv
    path: data/orders.csv

transform:
  monthly:
    from: orders
    group_by: [region, month]
    aggregate:
      revenue: sum(amount)

output:
  monthly:
    type: parquet
    path: build/monthly.parquet

The full grammar, semantics, and connector reference live in the documentation.

Two artifacts

BaruchML: The specification. Versioned on its own, and what the documentation tracks.
BaruchAPI: The Python reference implementation, on PyPI as baruch.