AI-Integrated Jewelry E-Commerce Platform: A Full-Stack Web Application with Generative Design, Conversational AI, and Real-Time Market Intelligence

AI-Integrated Jewelry E-Commerce Platform: A Full-Stack Web Application with Generative Design, Conversational AI, and Real-Time Market Intelligence

Harsh Mali¹, Arsh Naikawadi², Aryan Kadam³, Om Salunkhe-Patil⁴, Sahil Jagtap⁵
¹ Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Nagthane, Sangli, Maharashtra, India, er.maliharsh95@gmail.com

² Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Dapoli, Ratnagiri, India, arshnaikawadi@gmail.com

³ Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Karad, Satara, Maharashtra, India, aryankadam2706@gmail.com

⁴ Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Nagthane, Sangli, Maharashtra, India, ompatil151107@gmail.com

⁵ Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Mitharwadi, Kolhapur, Maharashtra, India, sahiljagtap1123@gmail.com

Guide: Mrs.Madhuri Kamble, Dept. of Computer Engineering, Nanasaheb Mahadik Polytechnic Institute Peth, A/P: Walwa, Sangli, Maharashtra, India, madhurikamble848@gmail.com

ABSTRACT
The Indian jewelry retail sector, valued at over USD 85 billion, continues to rely on conventional storefronts with limited digital engagement, resulting in poor customer reach, opaque pricing, and absence of personalized design capabilities. This paper presents the design, implementation, and evaluation of a full-stack, AI-integrated web application developed for Amar Jewellers, a heritage jewelry retailer based in Sangli, Maharashtra. The proposed system employs a modern three-tier architecture comprised of a React-TypeScript single-page application (SPA) on the frontend, an Express.js RESTful API server on the backend, and MongoDB as the persistence layer. The platform introduces three novel capabili- ties for traditional jewelry retailers: (1) an AI-powered Design Studio that leverages the Gemma 3 large language model via the Bytez inference API to generate custom jewelry visualizations from user-specified parameters such as metal type, style, weight, and budget; (2) an intelligent conversational chatbot named "Amar" that provides context-aware re- sponses including live commodity rates and in-page navigation; and (3) real-time gold and silver price feeds sourced from the Amar Bullion broadcast streaming API. Additional modules include a Google Sheets-integrated repair tracking system with unique ticket identifiers, JWT-based multi-role authentication (customer and administrator), and a curated product catalog with dynamic category filtering. The user interface incorporates Framer Motion-driven micro-anima- tions, Lenis smooth scrolling, and glassmorphism aesthetics to deliver a premium digital experience. Experimental eval- uation demonstrates sub-second API response latencies for rate fetching, seamless AI design generation within accept- able timeframes, and a fully functional repair lifecycle from submission through status tracking. The system effectively bridges the gap between heritage craftsmanship and modern digital retail by democratizing access to AI-driven jewelry design for small and medium-sized jewelry enterprises.

Keywords: AI-Powered Jewelry Design, E-Commerce Platform, React TypeScript, Conversational AI, Real-Time Market Data, Full-Stack Web Application, Generative AI, MongoDB, Express.js