"AstraScore: AI-Powered Alternative Credit Scoring System"

• 

  “Displays a real-time AstraScore using digital behavior, highlighting UPI discipline, spending history, AI-driven credit improvement tips.”

Inspiration

In India and other emerging economies, over 400 million adults are classified as “New-to-Credit” (NTC), making them invisible to traditional credit scoring models. Banks depend heavily on CIBIL-like reports, which do not reflect the financial discipline of individuals without prior loans or credit cards. To overcome this problem AstraScore, an AI-powered alternative credit scoring system that evaluates risk

What it does

This project proposes AstraScore, an AI-powered alternative credit scoring system that evaluates risk using non-traditional signals such as UPI transaction behaviours, bill payments, savings patterns, recurring expenses, geolocation stability, mobile metadata, and digital behavioural signals. The solution uses machine learning classification models, supported by explainable AI (XAI), to generate a fair, transparent, and real-time credit score.

How we built it

We built AstraScore using alternative digital behavioural data, processed it with machine learning models, and deployed real-time, explainable credit scoring through a FastAPI-based backend.

Challenges we ran into

The main challenges included limited access to real financial data, ensuring user privacy, handling noisy behavioural signals, and maintaining fairness and transparency in AI predictions.

Accomplishments that we're proud of

We built a working AI-based alternative credit scoring system that generates real-time, fair, and explainable credit scores for New-to-Credit users, enabling their creditworthiness to be accurately reflected and improving access to loans and financial opportunities.

What we learned

What's next for Untitled

Built With

• java
• javascript
• json
• python
• react
• sql
• yaml