Labeler
Labeled and categorized thousands of bank transaction records for a consumer lending client. Each transaction was tagged with merchant category codes (e.g., grocery, dining, utilities, entertainment) to help train a gradient-boosted model that predicted loan default risk. Worked with ambiguous transactions (e.g., "POS PURCHASE" without merchant details) and applied consistent logic to assign correct categories. Performed quality control by reviewing label consistency across multiple batches and resolving edge cases with the client's guidelines.