🏦 Banks & Scheduled Commercial Banks

UPI Fraud Detection Built for Indian Banks

Purpose-built for the regulatory obligations, transaction volumes, and operational workflows of scheduled commercial banks. Not a generic platform adapted for India — designed for it.

Request Bank Demo RBI Compliance →
<50ms
Transaction scoring latency
16B+
UPI transactions processed monthly in India
6h
CERT-In incident report deadline — met automatically
20
RBI compliance checkpoints covered

What vcurd solves for banks

The four fraud and compliance challenges where Indian banks need the most support — and where generic platforms fall short.

Real-Time UPI Transaction Blocking

Rule engine + ML pipeline runs in under 50ms. Block high-risk transactions before UPI completion — not after. Hard BLOCK, soft FLAG, step-up 2FA, and DELAY actions all configurable without code changes. Essential for meeting the RBI real-time monitoring mandate.

🕸️

Mule Account Network Detection

Graph-based analysis of the full VPA→VPA transaction network. Computes composite Mule Score across 6 signals: account dormancy, pass-through ratio, device sharing, VPA age, MHA district, and SEON intelligence. Catches 2nd and 3rd-hop mules that rule-based systems miss.

📋

RBI FMR & CERT-In Compliance

One-click FMR export in the RBI-required format. Automated CERT-In incident report generation. Compliance countdown dashboard with deadline tracking. Full immutable audit trail. SAR/STR filing workflow for FIU-India. Everything needed for examination readiness, always available.

📊

Card BIN Intelligence

Full card network, issuing bank, card type, and international flag for every card transaction. Country-level fraud pattern analysis. Separate card fraud rules distinct from UPI rules. Real-time card fraud heatmap by network and transaction type — critical as card-not-present fraud scales.

How a transaction flows through vcurd

Transaction received at API endpoint

Every UPI transaction hits the scoring API. Full transaction context passed: sender VPA, receiver VPA, amount, device ID, location, IP, card BIN if applicable, and behavioral flags.

Rule engine evaluation (<5ms)

Enabled rules evaluated in priority order. First BLOCK terminates the pipeline — sub-5ms for pure rule-based blocks. Conditions evaluate amount thresholds, velocity, time-of-day, device flags, VPA blacklist, and 12+ other signals.

ML model scoring (10–40ms)

XGBoost + Isolation Forest pipeline runs if rule engine does not BLOCK. 23 features including engineered signals specific to Indian UPI fraud. Score 0–1 compared to configurable fraud threshold.

Decision & SHAP explanation

Final decision (BLOCK / FLAG / DELAY / 2FA / PASS) returned with SHAP attribution showing which features drove the score. Every decision logged to immutable audit trail.

Alert & WebSocket broadcast

Flagged transactions create Alerts visible to risk analysts in the live dashboard. WebSocket push to all connected clients means the operations team sees fraud events in real time — not on a T+1 report.

Modules most used by banks

Real-Time ML Scoring
⚙️Dynamic Rule Engine
🕸️Mule Account Detection
💳Card BIN Intelligence
🗺️Geographic Fraud Heatmap
📋RBI Compliance Reports
📑SAR / STR Filing
🔔Alert Management
🧾Bureau Velocity
🤖Swarm Agent Analytics
📁Case Management
📡Live WebSocket Feed

Built for your bank's fraud operations team

Schedule a demo with a use case specific to your institution's fraud profile.