data-pipelinelisted

# Data Pipeline 1. Confirm source ownership, permitted use, availability timing, and storage boundary. 2. Keep inputs outside version control or in the project's explicitly ignored private area. 3. Define grain, primary keys, types, null meaning, and expected coverage before transforms. 4. Validate row counts, uniqueness, joins, non-matches, ranges, and freshness at each boundary. 5. Use DuckDB or Polars for large columnar workloads when justified. 6. Write code and schema contracts, never private data or derived datasets. 7. Record evidence and limits using the project's durable-output contract; do not fill unavailable values silently.