I turn messy, inconsistent data into clean, validated, de-duplicated CSV or JSON files, and build the automation around it. I extract data from websites, PDFs, spreadsheets and documents, then normalize the fields into a consistent schema such as product name, brand, price, currency, category and stock. Every row is validated with a confidence score, and uncertain rows are flagged for human review. My pipelines are idempotent, so re-runs never create duplicates, and every delivery ships with a short QC report covering coverage and flags. I work in Python with requests, Playwright, Pandas, SQL and ETL, and my AI extraction is checked by real validation rules rather than an LLM simply guessing. Best fit is fixed-price projects with milestone escrow. Native Korean, fluent English, with fast and clear communication.