API Reference Support GitHub →

Best Practices

Get the most out of the AI Body Scan API with these recommended patterns and tips.

🔑 API Key Management

🔒

Secure Storage

  • Store API keys in environment variables, never in code
  • Use secrets management (AWS Secrets Manager, GCP Secret Manager)
  • Never commit keys to version control
  • Rotate keys regularly (every 90 days)
🚫

Key Scoping

  • Use separate keys per environment (dev/staging/prod)
  • Implement proper key prefixes (sk_live_, sk_test_)
  • Set up separate keys per application
  • Monitor key usage for anomalies

📸 Photo Quality

✅

Pre-validation

  • Validate image dimensions client-side before upload
  • Check file size (max 16MB)
  • Verify image format (PNG, JPG, WebP)
  • Show preview before submitting
📱

Mobile Optimization

  • Compress images before upload (>1MB)
  • Use progressive JPEG for faster upload start
  • Implement offline queue for poor connectivity
  • Show upload progress indicator

⚡ Performance

⏱️

Request Optimization

  • Use connection pooling for highvolume
  • Implement request timeout (30s recommended)
  • Add retry with exponential backoff
  • Cache health check results (5 min TTL)
📊

Rate Limiting

  • Implement client-side rate limiting
  • Queue requests during high traffic
  • Use webhooks for async processing
  • Monitor 429 responses

🛡️ Error Handling

❌

Retry Strategy

  • Retry on 5xx errors (exponential backoff)
  • Don't retry on 4xx errors
  • Max 3 retries per request
  • Log failures for monitoring
💬

User Feedback

  • Show specific error messages
  • Provide actionable suggestions
  • Allow easy photo retake
  • Don't expose raw API errors

📈 Monitoring

📉

Key Metrics

  • API latency (p50, p95, p99)
  • Success/failure rates
  • Measurement accuracy feedback
  • Quota usage tracking
🔔

Alerting

  • Alert on >5% error rate
  • Alert on quota >80%
  • Alert on high latency
  • Set up webhook notifications

🏗️ Architecture

🔄

Resiliency Patterns

  • Implement circuit breaker
  • Use bulkhead pattern for isolation
  • Add dead letter queue for failures
  • Plan for graceful degradation
📦

Scalability

  • Design for horizontal scaling
  • Use async processing
  • Implement caching layer
  • Consider CDN for static assets