SmartCampus AI: A Proactive, AI-Enhanced ERP for Reducing Student Disengagement in Public Higher Education

Team Name: Desert Coders

Team ID: 87867

Abstract Traditional Enterprise Resource Planning (ERP) systems in public educational institutions often function as passive, disconnected systems of record. This paper details the architecture and implementation of SmartCampus AI, a modern, full-stack, and integrated educational platform designed to address this gap. The system consolidates core campus operations—including admissions, academics, and hostel management—into a single source of truth and introduces a key innovation: a Predictive Intervention Engine. By leveraging a Python-based Machine Learning model to analyze real-time student data such as attendance, marks, and fee status, the platform proactively identifies at-risk students and presents this analysis to mentors on an intuitive dashboard. This shifts the paradigm from reactive problem-solving to proactive, data-driven student support, with the goal of reducing dropout rates in a financially sustainable manner for public colleges.

1. Introduction & Problem Statement Many public colleges in India operate with fragmented data systems, often relying on disparate spreadsheets for critical functions like attendance, examination records, and fee collection. This data siloing makes it nearly impossible to get a holistic view of a student's well-being and academic trajectory. Problems are often revealed by term-end results, by which time struggling students may have already disengaged beyond effective intervention. The challenge is to create a low-cost, integrated system that empowers educators to act early. SmartCampus AI was built to solve this exact problem by creating a unified data ecosystem and layering predictive analytics on top of it.

2. System Architecture & Technical Approach The platform is designed as a modern, scalable, and secure full-stack application.

Backend: A robust REST API built with Java and the Spring Boot framework. It follows a three-tier architecture (Controller, Service, Repository) for a clean separation of concerns. Security is managed by Spring Security with JWT for stateless, role-based authentication and authorization.

Frontend: A dynamic single-page application (SPA) built with JavaScript and React (using Vite). The UI is composed of reusable components and uses a global Auth Context to manage user sessions.

Database: A centralized PostgreSQL database serves as the single source of truth for all modules.

AI Microservice: A lightweight Python microservice using the Flask framework hosts the machine learning model. This decoupled approach allows the AI engine to be scaled and updated independently of the core ERP application.

3. Core Feature: The Predictive Intervention Engine The most innovative component of the platform is the AI-powered risk analysis system.

Data Aggregation: The Java RiskAnalysisService gathers real-time data points from the central database for each student assigned to a mentor. This includes overall attendance percentage, the average score from the most recent examination, and the status of tuition fee payment.

Prediction: This data is sent via a REST API call to the Python microservice. The service uses a pre-trained Logistic Regression model from the Scikit-learn library to analyze the input features.

Output: The model returns a risk probability score between 0.0 and 1.0, indicating the likelihood that a student is at risk of disengagement or failure.

Visualization: The React frontend consumes this data and displays it on the mentor's Student Risk Dashboard. Each student is presented with a color-coded card (High, Medium, or Low risk) and a percentage score, allowing the mentor to instantly identify who needs attention. Clicking on a student reveals the specific contributing factors, such as "Attendance: 65%".

4. Feasibility and Impact

Sustainability: By building exclusively on powerful open-source technologies, the platform eliminates high licensing fees associated with commercial ERPs. The proposed deployment stack (Vercel, Render, Supabase) has a generous free tier, making the solution highly affordable and sustainable for public colleges.

Social Impact: The primary impact of this system is the potential to significantly reduce student dropout rates. By providing mentors with early, data-driven insights, the platform facilitates timely intervention. This democratizes student support and ensures that every student has a better chance to succeed, empowering educators and transforming the institution from a reactive to a proactive force in its students' lives.

5. Future Work The modular architecture of SmartCampus AI allows for significant future expansion. The immediate next steps include:

Building out the automated notification system to alert mentors via email when a student is flagged as high-risk.

Implementing a full-featured Library Management module.

Integrating a payment gateway to automate the collection of fees and the generation of digital receipts.