Apparel Classifier Demo

A live web app that uses an LLM to recover corrupted retail product classification labels, demonstrating AI-powered master data repair.

URLs

• Frontend (CloudFront): set after infra/bootstrap.sh runs — record the distribution domain.
• Backend (API Gateway): set after infra/bootstrap.sh runs — record the invoke URL.

Local development

See CLAUDE.md for the canonical run/test/deps instructions. In short: backend in backend/ (FastAPI + uvicorn on :8000), frontend in frontend/ (Vite dev server on :5173 with /api proxied to the backend).

Secrets

.env.example lists every key required at deploy time. Real values come from Eve-Hwang org GitHub Actions secrets; nothing is committed to the repo.