CDK Starters

API Gateway + Lambda + DynamoDB with TypeScript. Full CRUD boilerplate with IAM roles, environment config, and local testing setup.

CloudFront distribution with S3 origin and Origin Access Control, HTTPS redirect, security headers, SPA error routing, and optional custom domain with ACM + Route53.

EventBridge scheduled rule triggering a Lambda function on a cron or rate schedule. Includes DynamoDB state table, CloudWatch error/duration alarms, and SNS alarm notifications.

SNS fan-out to three SQS queues (notify, analytics, audit) each with a Lambda consumer and dead-letter queue. Includes a custom EventBridge bus for cross-service domain events and DLQ alarms.

Production-ready VPC with 3 AZs, public/private/isolated subnets, NAT gateways, S3/DynamoDB gateway endpoints, ECR/SSM/CloudWatch interface endpoints, VPC Flow Logs to S3, and an SSM-only bastion host.

Cognito User Pool with hosted UI, SPA and server-side app clients, and an API Gateway REST API protected by a Cognito JWT authorizer. Includes public and protected Lambda endpoints.

CloudFront + S3 frontend, API Gateway + Lambda backend, Cognito auth, and Aurora Serverless v2 PostgreSQL. Production-ready full-stack pattern with all wiring included.

Containerised service on ECS Fargate behind an Application Load Balancer. Includes ECR repository, ARM64 tasks, CPU + request-count auto-scaling, health checks, and CloudWatch alarms.

Self-mutating CDK Pipeline connected to a GitHub repository via CodeStar Connection. Deploys through Staging and Production stages with manual approval and a smoke test step.

API Gateway WebSocket API with Lambda handlers for $connect, $disconnect, and custom message routes. DynamoDB connections table with userId GSI for server-push messaging.

Express Step Functions workflow with validate → process → notify steps, error handling with Catch/Retry, Choice branching, and an API Gateway trigger endpoint. CloudWatch failure alarm included.

Conversational AI chatbot powered by Amazon Bedrock (Claude 3.5 Sonnet). Lambda handler with multi-turn conversation history in DynamoDB (TTL-based). REST API with session management.

S3-based data lake with raw/processed/curated zones, Glue catalog and daily crawler, Athena WorkGroup, and a Kinesis Data Stream + Firehose ingestion pipeline. ETL Lambda for raw-to-processed transformation.

Temporal workflow server on ECS Fargate — no Kubernetes or Helm required. Aurora PostgreSQL 15 Serverless v2, NLB on port 7233, Route53 DNS, and optional Web UI sidecar. Simpler and typically cheaper than the standard EKS/Helm approach.