VyuhaDB: An open-source repository for structured Indic-language scam signatures to empower AI-driven fraud detection in India. #AIforAll #AIKosh #TeamAbhaya

Team: Abhaya
Theme: AI for Social Impact / Data Readiness
Focus: Protecting Senior Citizens and MSMEs from Cyber-Fraud

---

Project Overview

VyuhaDB is an automated open-source pipeline and repository designed to bridge the "AI-Readiness Gap" in Indian cybersecurity. We transform unstructured fraud alerts and news reports into a structured "Signature Library" of scam scripts in regional languages (Malayalam, Hindi, and English).

Inspired by the strategic defensive formations of ancient India, VyuhaDB organizes chaotic data into a machine-readable defense shield for the national AI ecosystem.

The Problem

India loses thousands of crores annually to localized "Social Engineering" scams. Global AI safety tools fail because they lack high-quality, structured datasets of Indic-language scam scripts (Manglish/Hinglish).

The Solution

VyuhaDB extracts the "Linguistic DNA" of scams.

1. Automated Ingestion: Scrapes raw data from official portals (NCRP, Sanchar Saathi).
2. Indic Parser: Uses Bhashini APIs to standardize regional scam scripts.
3. Signature Registry: Provides a JSON/CSV library of scam patterns for developers to build "Safe-Dialers" and fraud-blockers.

Data Schema (Vyuha-Signature)

The repository provides signatures in a standardized format:

{
  "scam_type": "Digital Arrest",
  "language": "Malayalam-English",
  "psychological_trigger": "Fear/Authority",
  "script_snippet": "CBI office ninnu vilikkunnu... Aadhaar illegal activity undu...",
  "risk_level": "High"
}