HOMEPROJECTSBLOGJOURNEYSKILLSCONTACT
Fazrin Nugraha
FAZRINPORTFOLIO
Navigation
HOMEPROJECTSBLOGJOURNEYSKILLSCONTACT
© 2025 Fazrin Portfolio
← Back to Home
October 2025CASE STUDY

SocialCollateral AI

Specialized Social Graph Engine for Grameen-style Trust Scoring

About The Project

SocialCollateral AI is a specialized Social Graph Engine developed to address a critical paradox in the microfinance industry: how to scale inclusive lending without losing the “social touch” that mitigates risk. As microfinance lenders like Amartha expand, they face an “asymmetry of information” where transaction data exists, yet the invisible dynamics of communal trust—the very foundation of the Grameen model—remain unquantified and “blind” to digital systems.

This project introduces “Jaringan Amanah”, a system that transforms these qualitative social assets into a quantitative Group Trust Score. By shifting the paradigm from individual credit scoring to graph-based risk assessment, we enable financial institutions to “see” the strength of social fabric and monetize the reputation of borrower groups.

The solution is architected around a Social Graph Engine that triangulates risk through Three AI Lenses:

  1. Network Topology (Graph Analytics): Visualizing and computing the structural integrity of borrower clusters using centrality metrics to identify “toxic” versus “solid” groups.
  2. Contextual Sentiment (NLP): Utilizing Google Gemini to decode unstructured field agent reports, detecting nuanced signals of conflict or cohesion that purely numerical data misses.
  3. Physical Asset Valuation (Computer Vision): Deploying Google Cloud Vision to objectify the assessment of business and household assets, reducing human survey bias.

This integrated approach enables Precision Lending (reducing opportunity loss and bad debt) and Operational Velocity (automating pre-screening), effectively turning the “invisible” collateral of social trust into a bankable asset for the unbanked.

Built With

  • TypeScript
  • React
  • TailwindCSS
  • Sigma.js
  • Chart.js
  • Python
  • FastAPI
  • Google Gemini
  • Google Cloud Vision

Contributors

Our team built this complete system in a 24-Hour National Hackathon to secure the Finalist Top 15 finish.

  • Muhamad Fazrin Nugraha - Frontend Developer & Interface Design
  • Collaborators - Data Scientists, ML Engineers & Backend Developers

Acknowledgments & Inquiries

Special thanks to Amartha for providing the challenge context and fostering innovations in fintech and microfinance inclusion.

For any inquiries or feedback regarding the system architecture or interface design, please contact nugrahafadzrin@gmail.com.

Project Case Study Showcase · Muhamad Fazrin Nugraha