Projects & Delivery

End-to-end email flow with SES rulesets, S3 storage, Lambda processing, and reliable fan-out via SQS plus observability baked in.

• SES inbound delivers to S3 prefix; S3 event triggers Lambda to parse metadata and push work messages onto SQS while fetching raw content for processing.
• Lambda workers pull from SQS, perform enrichment/operations, and send SES outbound responses; OTEL collector layer streams traces to X-Ray and metrics/logs to CloudWatch.
• CloudWatch alarms for delivery failures/DLQ depth; at-rest encryption and least-privilege IAM across buckets, queues, and functions.

Built a secure, multi-tenant delivery layer with CloudFront OAC and signed URLs backed by API Gateway + Lambda (TypeScript), plus AI workflows for transcription and translation.

• OAC-protected S3 origins, per-tenant routing via Route53, and signed cookies/URLs for uploads and downloads.
• Bedrock + Transcribe + Comprehend + Translate pipelines for multilingual care-team communication.
• End-to-end observability with CloudWatch dashboards/alarms, X-Ray tracing, and Lambda alert handlers.
• Terraform/Serverless modules for repeatable deployments across environments.

Zero-public-access ML platform with strict network boundaries and authenticated inference across accounts.

• NLB → ALB mTLS with ACM PCA-issued certs; exposed privately via PrivateLink and Direct Connect.
• Lambda layers for JWT/JWS validation, request transformation, and image preprocessing (distroless).
• ADOT + CloudWatch logs/metrics/alarms and encrypted S3 audit trails for compliance.

Real-time ingestion and decision support for airports, combining weather feeds with flight ops data.

• Lambda/EventBridge ingests Tomorrow.io schedules and events into MSK.
• EKS microservices merge weather, runway visibility, and flight schedules for alerting and dashboards.
• Prometheus + CloudWatch + OTEL tracing for full path visibility.

Python automation to reclaim waste and enforce retention while keeping compliance intact.

• Cleanup for unused EBS volumes, stale AMIs/snapshots, and outdated RDS snapshots with guardrails.
• Scheduled EC2 shutdowns for non-prod; enforced CloudWatch log retention policies.
• Delivered recurring monthly savings across compute, storage, and logging.

Modernized brittle Bash+CloudFormation pipelines into Serverless Framework + Compose.

• Service separation with declarative deploys; parallelized environments via Compose.
• Removed “nuke and pray” rebuilds; improved dependency hygiene and rollback confidence.
• Cut deployment times by ~50% and reduced operational overhead.