System Requirements
- Classify every incoming email before delivering to inbox in < 100ms
- Support high-throughput stream ingestion for incoming SMTP traffic
- Rule-based fallback engine when ML inference models are degraded
- User feedback telemetry loop (spam/not-spam clicks) to retrain models
- System metrics, quarantine storage, and error alerting
Graded Constraints
01Total estimated architecture cost must stay under $3,000/month
02Must include Client / Ingestion source, Load Balancer or Gateway, and Compute Service
03Must connect a Message Queue / Stream (Kafka or Queue) for asynchronous feedback processing
04Must include Persistent Storage or Database for quarantine & classification rules
05Compute engine must have both input from gateway and output to storage/queues
Ready to design?
Open this challenge in the interactive system design playground and satisfy all architecture constraints.
Explore More
All ProblemsRelated Architecture Challenges
BitlyEasy
URL Shortener
Build a high-performance URL shortening service like TinyURL or Bitly that generates 7-character aliases, handles 100:1 read-to-write ratios, and caches hot redirect lookups.
100M URLs/month →
CloudflareEasy
Distributed Rate Limiter
Design an edge-tier rate limiting system to protect APIs against abuse, credential stuffing, and DDoS using sliding-window counter algorithms and fast distributed caches.
50,000 requests/sec →
UberMedium
Scalable Notification System
Architect a multi-channel notification engine supporting iOS APNS, Android FCM, SMS (Twilio), and transactional emails with priority routing and deduplication.
10M+ notifications/day →