Reference platform

Architecture

A complete cloud platform, layer by layer. Scroll to take it apart, hover a block to see how we build it.

Scroll · drag to rotate · hover a block

And yours?

Does your platform look like this?

In two weeks, we audit every layer of your infrastructure and hand you a prioritised action plan.

A typical architecture, the kind we design and run for clients: not any one client’s system.

Your browser can’t display 3D: here are the platform’s five layers.

  1. 01 · Network & security

    The private network, access and protections everything else stands on.

    Networks isolated per environment, least-privilege access, encryption in transit and at rest, all described in code.

    • Private networks — Split by environment and by availability zone.
    • Web application firewall — Stops web attacks before they reach the application.
    • Identity & access — Every person and service gets only the rights it needs.
    • Admin access — An encrypted tunnel, with no admin port exposed.

    Tools — Terraform, VPC, WAF, IAM, WireGuard

  2. 02 · CI/CD & infrastructure as code

    The chain that turns a commit into a production release, with no manual step.

    Pipelines tested on every commit, scanned images, infrastructure versioned and reviewed like code.

    • Git repository — Code and infrastructure in one place, with mandatory review.
    • CI pipeline — Build, tests and security scans on every commit.
    • Image registry — Versioned images, scanned before any deployment.
    • Infrastructure as code — Every cloud resource described, versioned and reproducible.

    Tools — GitHub Actions, GitLab CI, Docker, Terraform, Argo CD

  3. 03 · Runtime & Kubernetes

    Where your applications run: they restart on their own and follow the load.

    Managed clusters, progressive rollouts with rollback, autoscaling, encrypted secrets.

    • Kubernetes cluster — Several zones, to survive a data-centre outage.
    • Ingress & certificates — Traffic arrives encrypted and spread over healthy instances.
    • Autoscaling — More capacity at peak, less at night: you pay for real use.
    • Secrets — Passwords and keys encrypted, never in the code.

    Tools — Kubernetes, Helm, Argo CD, KEDA, Vault

  4. 04 · Data platform

    The company’s data in one place, reliable and ready for BI and AI.

    Streaming and batch ingestion, a modelled warehouse, dashboards, backups that are restored and tested.

    • Event streams — Data from applications and machines arrives continuously.
    • Data warehouse — One documented model that everyone can query.
    • Dashboards — The company’s KPIs, up to date with no manual export.
    • Backups — Automatic, encrypted, and restored regularly to prove it.

    Tools — PostgreSQL, Kafka, dbt, Power BI, Grafana

  5. 05 · Observability

    See what is happening, get warned before your customers, understand fast.

    Correlated metrics, logs and traces, alerts on what users experience, written on-call runbooks.

    • Metrics — Latency, errors, saturation: system health in numbers.
    • Logs — Every log in one place, searchable in seconds.
    • Traces — The path of one request through every service.
    • Alerts & on-call — A useful alert, a written runbook, a person notified.

    Tools — Prometheus, Grafana, Loki, OpenTelemetry, Alertmanager