// project

02 / 06

FitIQ

Apparel ranking system for surfacing products likely to satisfy users on fit and quality.

FitIQ ranks apparel products inside comparable groups using weak public signals from Amazon Reviews 2023. The pipeline includes review ingestion, metadata-driven taxonomy building, product-level feature generation, leakage-safe offline ranking evaluation, CatBoost ranking models, a query-group hybrid serving policy, a FastAPI service, and a Tkinter desktop demo client.