AI Credit Eligibility System

Admin Dashboard
Customers Loan Application
Admin Login Page
Bank Admin Dashboard
Loan Analysis Result By Ml Prediction
QR CODE FOR REPORT DOWNLOAD
Active Loans
Customer Profile
Customer Dashboard and chatbot interface
Loan Form
Customer Raising Ticket
Admin Resolving Ticket
Loan Amount to Be Transferred
Admin Bank Transfer Details
Admin Loan Transfer
Admin Customer Review

Inspiration

The inspiration for this project came from the realization that traditional bank loan processes are often opaque, slow, and intimidating for users. I wanted to build a transparent, "AI-First" lending platform that doesn't just give a "Yes" or "No," but actually explains the logic behind the decision using modern data science, making financial literacy accessible to everyone.

What it does

The AI Credit Eligibility System is a full-stack banking application that provides: Instant AI Decisions: Uses XGBoost and Random Forest models to predict loan approval probability in seconds. Explainable AI (XAI): Integrates SHAP (SHapley Additive exPlanations) to show users exactly which factors (like income, debt-to-income ratio, or employment history) impacted their score. Interactive Dashboard: A premium UI for monitoring credit health, active loans, and financial risk profiles via radar and bar charts. Bank-Grade Reporting: Generates RBI-compliant PDF analysis reports with unique QR codes for secure mobile downloading. End-to-End Workflow: Includes secure JWT authentication, a comprehensive loan application form, and a simulated payment gateway.

How we built it

rontend: Built with React and TypeScript using Vite. Styling was done with Tailwind CSS, and data visualizations were implemented using Recharts and Lucide icons. Backend: Powered by FastAPI (Python) for high-performance asynchronous logic. Machine Learning: Developed using Scikit-learn, XGBoost, and Pandas. Used SHAP for generating feature importance values. Database: PostgreSQL (hosted on Neon DB) with SQLAlchemy for asynchronous ORM management. Reporting: FPDF2 for dynamic PDF generation and QR Code libraries for mobile sharing.

Challenges we ran into

Unicode in PDF Generation: We faced significant issues rendering special characters (like the Rupee symbol ₹) in PDFs, which required custom font mapping and character replacement logic. Mobile Performance: Rendering complex AI visualizations (Radar charts and SHAP bars) caused hangs on older mobile devices, leading us to optimize the "Analyzing" phase with better state management. Model Explainability: It was challenging to translate raw mathematical SHAP values into "Human-Readable" advice that a non-technical user could understand.

Accomplishments that we're proud of

High Model Accuracy: The Random Forest model achieved a 99.88% accuracy and a 1.00 ROC-AUC score on the processed evaluation dataset. Seamless Integration: Successfully connecting the complex ML inference logic to a real-time React dashboard without significant latency. The "AI Advisor" Experience: Creating a UI that feels premium and interactive, specifically the logic that suggests adding a co-applicant when a user is on the "borderline" of approval.

What we learned

The Power of XAI: Learned that in fintech, the explanation of a decision is often as important as the decision itself. Full-Stack Mastery: Deepened knowledge of asynchronous Python (FastAPI) and modern React patterns (Hooks, complex state management). Data Security: Learned how to handle sensitive financial records and implement secure "on-the-fly" report generation without unnecessary database bloat.

What's next for AI Credit Eligibility System

Bank Officer Dashboard: A dedicated portal for bank employees to manually review and approve "Pending Review" applications. Advanced Data Privacy: Finalizing the "Optional Persistence" feature to allow users to run analysis without saving their data to the cloud. Real-time Document Verification: Integrating OCR to automatically scan and verify identity documents (KYC) and bank statements.

Customer Credentials : Mobile Number : 8095006741 Customer ID : LA20253834 Email: vinay@gmail.com Password: Vinay@123 Pin: 234124 Link of Customer Page: https://ai-loan-advisor-three.vercel.app/

Admin Credentials : Admin ID: LAAD202501 Email: Vinaykumarsm2341@gmail.com ( case sensitive use exact) Password: Vinay@123 Pin: 234124 Link of Admin page: https://ai-loan-advisor-uaoz.vercel.app/login