Systems and Cloud Solutions Architect

The role

The work is primarily SRE and platform engineering — running production Kubernetes, owning the GitOps delivery path, automating infrastructure with Terraform, and operating the satellite-data pipelines that ICIPE's research depends on. I also ship application code on the platforms I run. The work spans multiple internal systems; Vector Atlas is the example I can point to publicly because it's open source.

Scope of work

• Kubernetes platform operations.Run production clusters that host ICIPE's research and data services — cluster lifecycle, workload topology, ingress/Gateway API, secrets and workload identity, rolling deploys, and incident response.
• GitOps with ArgoCD (App-of-Apps). ArgoCD is the primary GitOps controller I run; Flux is in use on adjacent projects. Bootstrap manifests track parent applications in Git; child applications are reconciled from chart and manifest repositories without manual kubectl.
• Infrastructure as Code with Terraform. Cloud resources, IAM, and cluster-adjacent infrastructure provisioned and versioned as code, with state managed for safe collaboration.
• CI/CD on GitHub Actions.PR CI, container image build & push, secret syncing to clusters, and automated image-tag updates that ArgoCD reconciles.
• Satellite-data pipelines.Cloud Optimized GeoTIFFs (COGs) live on S3 and are indexed in place via Open Data Cube; Argo Workflows orchestrates ingestion and processing jobs against that data on Kubernetes — no bulk transfer, no duplicated state. Results are published through tile servers and OGC web services.
• Software engineering on the application stack. Ship features across TypeScript / Next.js frontends, NestJS / GraphQL APIs, and Python ingestion / back-end services running on the platforms.
• Vector Atlas (open-source example).One of the platforms I run, and the one I can point to publicly. Author and maintain the Helm chart and ArgoCD application that deploy its full service topology — UI, API, ingestion API, tileserver, Python backend, analytics, Redis, Postgres, and ingress — and ship application code across the UI, API, and deployment workflows. Live at vectoratlas.icipe.org.

Stack

Public repositories

Some of the platform work is open-source. The links below point to the public Vector Atlas application code and its Helm chart / ArgoCD configuration.

Reliability and operational practice

Deployments are rolling updates reconciled by ArgoCD; secrets live in the cluster (never in Git) and are surfaced through Helm values and workload-identity webhooks. Image tags are bumped by automated workflows and reconciled on the next sync. Stateful workloads (Postgres, Redis) are isolated from stateless application pods, with storage and lifecycle managed separately so stateless rollouts stay fast and safe.