Developer Platform & GitOps

→Multi-account AWS EKS with ArgoCD and GitHub Actions for 20+ microservices — 40+ engineers get a self-service path to deploy by merging a PR.

Secrets Automation

→AI secrets-provisioning workflow built with Security: scoped IRSA policies and Secrets Store CSI manifests, so teams self-serve secrets in minutes with no secret material in Git.

Networking & Migration

→Led the EKS migration to IPv6-only clusters, removing subnet-size limits on pod density and eliminating per-address IPv4 charges — a DNS64/NAT64 path kept IPv4-only registries and GitHub working with zero application changes.

Observability & Reliability

→Replaced static CPU/memory alerts with Datadog APM, Prometheus, and Grafana SLO dashboards, paging only on real latency and error issues — cutting alert noise from ~35 to ~4 actionable pages a week.

Streaming Data Infrastructure

→Strimzi Kafka on EKS for cold-chain vaccine telemetry, partitioned across brokers for scale — load-tested to 2M+ daily events with no message loss observed.

Traffic Management

→Deployed the Kubernetes Gateway API on EKS to separate network management from app routing, letting developers safely update their own routes.

AI-Assisted Platform Work

→Claude Code skills that summarize failing pods, plus a Confluence MCP integration serving runbooks from the terminal — cutting repeat questions to the platform team.

Zero Trust Access

→Replaced static SSH keys and legacy VPNs with zero-trust access via Teleport across AWS and EKS, removing always-on production access and capturing full command audit logs, with an emergency access fallback.

FinOps & Cost Visibility

→Kubecost with Slack reports showing each team's spend — cutting non-production compute spend 30% over two quarters.

Disaster Recovery

→Velero daily EKS snapshots to cross-region S3, tested quarterly for under-15-minute production recovery.

Delivery Automation

→Replaced ad-hoc Python/Bash provisioning with Terraform and FluxCD, adding nightly drift detection so manual changes surfaced as alerts instead of outages.

CI at Scale

→GitLab CI pipelines with auto-scaling Kubernetes runners on AWS, giving developers self-service test and build workflows on every pull request.