Back to Projects
Case Study — 01

ZYRO Jewel Box

React.jsNode.jsMongoDBGemini AIGroq
ZYRO Jewel Box

System Architecture

ZYRO is built on a scalable MERN stack, deliberately optimized for high-performance e-commerce. The frontend utilizes React 18 with Vite, implementing the Intersection Observer API for silky 60fps infinite scrolling. The backend acts as an orchestration layer, integrating Google Gemini and Groq (Llama 3) for the AI styling engine, while Brevo handles transactional emails. MongoDB handles complex SKU-level inventory with selective data projection to keep API payloads minimal.

Technical Challenges & Solutions

Automated Order Status Management

Manually tracking and updating order statuses is prone to human error and creates bottlenecks at scale. I engineered a Node.js cron job that executes precisely at the top of every hour to automatically evaluate timestamps and transition order statuses across the database, ensuring seamless operational flow without manual intervention.

AI Virtual Try-On Pipeline

Coordinating two different AI models seamlessly. I used Groq (Llama 3) for rapid analysis of user portraits (skin undertones) and piped that context into Google Gemini to generate custom visualization composites in real-time.

Loyalty State Management

Built a custom ZYRO Coin System where users earn 10% back. Engineered automated expiration and deduction logic tied directly to the order lifecycle, ensuring returns automatically reverse loyalty point balances.