FresherConnect: An AI-Augmented Real-Time Smart Campus Management System with 3D Galaxy Visualisation, Role-Based Access Control, and Progressive Web Application Architecture

Campus management in higher education institutions continues to suffer from operational fragmentation --- attendance systems divorced from academic dashboards, communication isolated to informal channels, and student welfare monitored reactively rather than proactively. This paper presents FresherConnect, a comprehensive, AI-augmented, real-time campus management system designed and implemented as a B.Tech final-year project at Shri Sai College of Engineering and Technology (SSCET), Bhadravati. The platform unifies student authentication, role-based access governance, academic record tracking, real-time communication, event management, peer marketplace, gamified engagement, and AI-driven insights into a single cohesive interface. The defining visual innovation is a three-dimensional interactive galaxy rendered via React Three Fiber (R3F) and Three.js, in which every enrolled student is represented as a glowing star node dynamically positioned by department, academic year, and live engagement score. A multi-layer role-based access control (RBAC) architecture spanning Firebase security rules, Firebase Auth custom claims, Express.js middleware, and React component gates enforces five distinct roles --- Super Admin, HOD, Event Manager, Faculty, and Student --- with 100% gate accuracy in penetration testing. The AI layer delivers mentorship matchmaking via in-browser semantic embeddings (Xenova/all-MiniLM-L6-v2), a predictive at-risk student detector combining attendance, grade, and activity heuristics, HMAC-signed QR code geofence attendance, an XP-based gamification engine with streak tracking, and a Retrieval-Augmented Generation (RAG) campus assistant grounded in institutional documents. The system is delivered as a Progressive Web Application (PWA) with a Workbox service worker, offline capability, and Firebase Cloud Messaging push notifications. Evaluation results demonstrate a P95 API latency of 187 ms under 150 rps load, a System Usability Scale (SUS) score of 84.2 ("Excellent"), and a RAG accuracy of 91.2% versus 54.8% for the non-RAG baseline.