A commerce platform with semantic search for devotional brassware
A full e-commerce build — catalogue, taxonomy, occasions, cart, wishlist, orders and payments — with embedding-based search so customers can find pieces by describing them rather than by knowing the exact product name.
Client: Siddi Brass · Status: live
The challenge
Devotional brassware is hard to search with keywords. Customers look for a deity, an occasion or a size, and often do not know the formal product name, so a conventional keyword search returns nothing and the sale is lost.
What we built
We built the platform on Next.js and FastAPI with a taxonomy that mirrors how customers actually shop — deity, occasion, category — and added embedding-based semantic search so a descriptive query finds the right piece. Payments run through Razorpay, media through Cloudinary, with a full admin surface for catalogue, pricing and content.
Engineering highlights
- Semantic product search over embeddings, alongside conventional filtering.
- Domain-specific taxonomy: deity and occasion are first-class, not tags bolted onto a generic catalogue.
- Razorpay payments, Cloudinary media with server-side cropping, Google Sign-In.
- Admin-managed price adjustments, site content and product media without developer involvement.
- Idempotent startup migrations so schema changes ship safely on every deploy.
- Sentry error monitoring and rate limiting, active only when configured.
Technology
Services used
Want something like this?
Tell us your problem and we'll tell you exactly how AI solves it and what it costs.