E-commerce

ThriftbyZee

ThriftbyZee is a fashion e-commerce platform specializing in curated thrift and vintage clothing collections, providing a seamless shopping experience with intelligent recommendations.

3 monthsLive Demo

The Problem

Thrift shopping online lacked the curated experience of physical vintage stores. Customers couldn't easily discover items that matched their style, and sellers had limited tools to showcase their collections.

The Solution

Built an e-commerce platform with curated collections, style-based recommendations, seller dashboards, and a streamlined checkout process optimized for fashion purchases.

Architecture

Next.js with App Router for optimal performance

Django REST API

PostgreSQL with full-text search

Image optimization pipeline

Recommendation engine based on purchase history

Technologies Used

Next.jsDjangoPostgreSQLRedisAWS S3Stripe

Challenges

1Building a recommendation system with limited initial data
2Optimizing image loading for fashion photography
3Creating a seller experience that rivals dedicated platforms

Engineering Trade-offs

1Chose hybrid full-text + vector search over full-text alone — better handles fuzzy style queries like 'boho summer dress', at the cost of maintaining an embeddings pipeline alongside the search index
2Used custom image loaders over a managed image CDN — kept infrastructure costs low at launch, at the cost of building and maintaining the optimization pipeline in-house
3Chose collaborative filtering over a more complex ML recommendation model — usable results from day one with limited data, at the cost of weaker recommendations for new users with no purchase history

Lessons Learned

1Fashion e-commerce is heavily visual — image quality matters more than features
2Seller onboarding is the biggest growth bottleneck
3Recommendation quality improves dramatically with more interaction data