Stop PII leaks before they hit your warehouse. By using dlt and Pydantic to enforce data contracts, you can sanitize or quarantine sensitive fields the moment they’re ingested.

dltHub

Four architectural patterns for enforcing PII governance at the data ingestion layer using dlt and Pydantic are presented: Sanitizer (hash/mask known PII fields), Gatekeeper (block entire batch on any violation), Guardian (allow schema evolution but scan new columns), and Quarantine (route bad records to a review queue instead of stopping the pipeline). Each pattern is backed by code examples and a runnable Marimo notebook, shifting PII enforcement from documentation into the CI/CD pipeline as enforceable data contracts.

How to protect PII with dlt and Pydantic