Self-initiated demonstration, not client work. Three Python pipelines pull auth-free public data over REST APIs and emit deduped CSVs in which every row carries the public source URL it came from, so any single row can be re-verified independently.

Snapshot counts the scripts actually returned: 1,056 pest-related citations across 829 NYC restaurants (NYC Open Data 43nn-pn8j, window 2026-08-07 to 2026-09-02); 5,041 debris-generating DOB build permits issued since 2026-08-07 (rbx6-tga4); 1,952 open postings across 28 live public Greenhouse/Lever boards, scanned 2026-09-07. These are point-in-time snapshots, not a live index, and no claim is made that any record converted into business for anyone.

Full case study, sanitized sample rows and honest limits: https://contra.com/p/rkorFQlE-nyc-public-data-lead-engine

Same pipeline as a scoped deliverable, Data pull in a day, $150: https://contra.com/products/1u2QatvL-data-pull-in-a-day