I clean one CSV using rules agreed before work starts. A typical small project covers up to 1,000 data rows, 20 columns and three exact cleanup or validation rules, plus exact duplicate removal.

You receive: • A reusable Python script and run guide • Cleaned data, with separate duplicate and review CSVs • An audit report that records changes and accounts for the input rows

I preserve the source file. Values that the agreed rules cannot resolve remain visible in the review file. I check row counts, rule behavior and edge cases before delivery. The script uses Python 3 and its standard library.

Before starting, we agree the rules, output columns, fixed price, deadline and revision limit. For this small scope, the target is three days after complete requirements arrive, with one revision within the agreed rules. Other file sizes, enrichment, fuzzy matching or external integrations need a separate quote.

The linked example is an original, self-initiated demo using synthetic records. I use AI tools to help develop the script, then run and check it. Data processing is local; your records do not need to be uploaded to an AI service.

Message me with a redacted sample, the desired result, duplicate criteria and deadline. Please have permission to share and process the data, and omit passwords, payment-card data, government IDs and regulated medical records.