Rollout plan and open decisions
This page keeps the rollout sequence and unresolved architecture choices visible without burying them in the main overview.
Practical starting scope for SPWHI
If SPWHI needs the most basic safe starting point, use this sequence:
Phase A - foundation
- create
spwhi-bootstrap - configure GitHub OIDC
- create Terraform remote state
- build the
spwhi-infrarepo skeleton - create
wardmitra-apiandwardmitra-uirepos
Phase B - web and data foundations
- provision VPC
- provision S3 and CloudFront for the web app
- provision RDS PostgreSQL in a controlled environment
Phase C - backend platform
- provision EKS
- move the Node.js API behind ALB
- introduce IRSA and baseline observability
Phase D - scaling and AI
- add Karpenter
- introduce async worker scaling
- isolate AI inference to dedicated capacity only when needed
This sequencing reduces cognitive load for the current team and delivers value without forcing every advanced decision up front.
Open decisions to resolve in later deliverables
These items are intentionally left as follow-up decisions:
- final Aviel DB migration path and cutover method
- MongoDB Atlas vs DocumentDB
- exact observability stack depth for production
- final GitOps model and ArgoCD rollout
- AI inference runtime choice and whether GPU nodes are needed immediately
- detailed security policy and secrets refresh behavior
Recommendation for the current team
The platform should grow in the same order the team can safely operate it:
- first make infrastructure repeatable
- then make releases predictable
- then add scaling and specialization
That order is more important than adopting every advanced AWS or Kubernetes feature immediately.