Present
- Own engineering standards for infrastructure as code, CI/CD, and production support across the client development team.
- Coach developers to handle routine DevOps tasks on their own, so delivery does not depend on one person.
- Add AWS SDK calls to application code and ship them through peer-reviewed pull requests.
- Operate a 7-service self-serve microservices platform on ECS Fargate across staging, release, and production — with Claude AI code review as a mandatory PR gate on every service.
- Built zero-downtime ECS pipelines: test gates, commit-SHA images, circuit-breaker rollback detection, and post-deploy ALB health checks.
- Moved all repositories to OIDC with branch-scoped least-privilege deploy roles — eliminating long-lived AWS credentials from CI/CD.
- Consolidated ETL infrastructure into one reusable Terraform module across 5 environments, with Multi-AZ RDS and deletion protection in production.
- Architected an elastic Fargate demo platform — 6 app instances from one parameterised Terraform config using
setproduct, mixing FARGATE and FARGATE_SPOT with CPU target-tracking and EventBridge Scheduler scale-to-zero. - Designed per-project Bedrock cost attribution: dedicated IAM identities, CUR 2.0 + Athena/Glue pipeline, daily CloudWatch cost dashboard, and 4-tier AWS Budgets alerts.
- Provisioned the AWS platform infrastructure (ECS Fargate, DynamoDB, S3, Secrets Manager, SSM Parameter Store) for a 10-portal vehicle-finance funder pricing service replacing a legacy RPA process.
- Provisioned Bedrock access to Anthropic Claude models in Terraform with EU inference profiles, per-project secrets, and keyless CI/CD access via STS external ID.
- Architected and delivered an MCP-based AWS investigation agent — 8 audited read-only tools, immutable audit log, automatic secret redaction, 20+ formal delivery phases, 1,689 passing tests.
- Deployed an AI compliance call-review service to production on ECS Fargate with full OpenTelemetry tracing and structured logging.