#10Real-Time Payment Fraud Detection (Radar)
10 / 20
Stripe logoStripe System
HardData Systems
40–60 min

Real-Time Payment Fraud Detection (Radar)

Design an ultra-low latency payment risk engine evaluating global network behavioral signals, velocity counters, and ML risk scoring before transaction authorization.

Scale & Traffic10,000 transactions/sec · < 80ms hard evaluation budget
Monthly Cloud Budget$30,000/month
Latency Target (P99)< 50ms Realtime SLA
Availability SLA99.99% Multi-AZ Active

System Requirements

  • Evaluate transactions against thousands of fraud rules in < 50ms
  • Global velocity counters (e.g., cards per IP in 10 mins)
  • Asynchronous risk graph construction without blocking payment flow
  • Zero-downtime rule updates and dynamic merchant risk thresholds

Graded Constraints

01Budget: $30,000/month
02API Gateway with strict TLS termination and mutual auth
03Dual path architecture: Synchronous decision path + Asynchronous feature pipeline
04In-memory ultra-fast cache for velocity checks
05Audit-proof immutable logging database

Ready to design?

Open this challenge in the interactive system design playground and satisfy all architecture constraints.

Explore More

Related Architecture Challenges

All Problems
Real-Time Payment Fraud Detection (Radar) — System Design Problem | Mimir Nest