Skip to main content

​
Deployment

Kaireon is containerized and deployed via Docker. Two images are built:
ImagePurpose
kaireon-apiMain Next.js application (UI + API routes)
kaireon-workerBackground job processing (pipelines, retraining)
Images are pushed to Amazon ECR and tagged with the git SHA plus latest.

​
Production Stack

ServiceProviderPurpose
App RuntimeAWS App RunnerHosts the Next.js application
DatabaseSupabase (PostgreSQL)Primary data store via Prisma 7
CacheUpstash (Redis)Enrichment caching, rate limiting
DNSRoute 53Domain management
CDNCloudFrontMarketing website + static assets
Container RegistryAmazon ECRDocker image storage

​
Monitoring

​
Prometheus Metrics

The platform exposes metrics at /api/v1/metrics:
  • kaireon_decisions_total — Total decisions made
  • kaireon_decision_latency_ms — Decision latency histogram
  • kaireon_pipeline_executions_total — Pipeline run counts
  • kaireon_api_requests_total — API request counts by endpoint

​
Grafana Dashboards

Pre-built Grafana dashboards are included in helm/dashboards/:
  • API Overview
  • Decision Engine Performance
  • Infrastructure Health
  • Model Health
  • Worker Queues

​
Helm Chart

A Helm chart is provided for Kubernetes deployment in helm/. Includes templates for:
  • API and worker deployments
  • ConfigMaps and secrets
  • Ingress with TLS
  • PodDisruptionBudgets
  • NetworkPolicies
  • Prometheus + Grafana stack

​
Rate Limiting

API endpoints are protected by a sliding-window rate limiter. Configurable per endpoint with Redis-backed counters.

​
Circuit Breakers

External integrations (connectors, webhooks) use circuit breaker patterns to prevent cascade failures. States: closed → open → half-open.