Project Setu
Project Setu is an adaptive, hybrid AI routing system engineered for microfinance institutions (MFIs) in emerging economies.
Core trick
Normalizes SMS logs, ledger photos, and voice notes into one schema, then routes each borrower to a scoring path matching the available inputs.
Description
Project Setu is an adaptive, hybrid AI routing system engineered for microfinance institutions (MFIs) in emerging economies. The platform empowers MFI field officers to instantly assess the creditworthiness of unbanked, informal-sector borrowers using messy, unstructured, and multi-modal inputs—such as SMS transaction logs, handwritten ledger photos (daybooks), and conversational voice notes.
README excerpt
# Project Setu: Adaptive AI Credit Risk Routing for Microfinance (MFI)
[](file:///d:/Desktop%20Data/ML/Projects/setu/Setu/project_overview.md)
[](file:///d:/Desktop%20Data/ML/Projects/setu/Setu/backend/main.py)
[](file:///d:/Desktop%20Data/ML/Projects/setu/Setu/frontend/package.json)
---
## 🌟 About the Project
**Project Setu** ("Setu" meaning *bridge* in Sanskrit) is an adaptive, hybrid AI routing system engineered for microfinance institutions (MFIs) in emerging economies. The platform empowers MFI field officers to instantly assess the creditworthiness of unbanked, informal-sector borrowers using messy, unstructured, and multi-modal inputs—such as SMS transaction logs, handwritten ledger photos (daybooks), and conversational voice notes.
Setu is built on a **hybrid Edge-Cloud architecture**:
- **Local Layer:** Lightweight, privacy-preserving, and offline-capable extraction of financial metrics from SMS records using a localized **Gemma 4 E4B** model running via `llama.cpp` on the edge.
- **Escalation Engine:** An anomaly-detection router that flags low-confidence or high-risk cases and seamlessly escalates them to cloud-hoste