← Tools

Python

transformation

Python for data engineering

Custom transformations, glue, and anywhere-code.

Python is the glue of every real data stack. Custom connectors, transformations that dbt can't express, ML pipelines, and the bash-script replacement layer. We write Python that's boring in the best way: tested, packaged, deployed, and owned.

Python · live signal

What we've shipped with Python

in production

0+

custom modules shipped

0%

typed + tested

0+

custom connectors

transformationopinionated · python

When Python makes sense

Pick Python when

  • You need a custom connector no vendor offers
  • Transformations need real logic (not just SQL)
  • You want typed, tested, reviewable data code
  • ML or scoring pipelines live alongside analytics

What we implement

Beyond the tutorial.

01

Custom connectors

Typed API clients with retries, pagination, and backfill.

02

Transformation modules

Pydantic-validated, tested modules used from dbt/Airflow/Dagster.

03

ML pipelines

Train/score loops with reproducibility, tracking, and deployment.

04

Package + deploy

Poetry / uv packaging, CI, and production rollout.

Not sure which fits?

30 minutes. We'll tell you honestlywhat's broken.