AI-Driven Multi-Source Credit Scoring

Traditional credit scoring often excludes underserved populations due to limited data sources (bank history, formal loans). In Vietnam and similar markets, millions remain “credit invisible” despite being active in digital ecosystems. How can we leverage unconventional data — SMS, e-commerce, social, and behavioral signals — with AI/ML to create fair, explainable, and inclusive credit scoring?

Expected Outcomes: - Develop an alternative credit scoring engine using unconventional data sources: SMS content, contact lists, social media, e-commerce activity, browsing trends.  - Apply AI/ML to ensure accurate, fair, and explainable credit scores.  - Integration prototype (e.g., simple API or scoring dashboard) for lender use - Bonus: integrate blockchain for secure, transparent, user-consent-based data handling.  Success Metrics: - Accuracy of creditworthiness prediction compared to traditional models - Demonstrated fairness and explainability (no strong bias against specific demographics)

None

None

- Underserved individuals lacking formal credit history - Financial institutions, neobanks, and fintech lenders seeking alternative risk assessment models

N/A

N/A