Architecture

Deploy faster. Spend less. Empower your developers. Kaja is a simplified way to develop and deploy—on your own infrastructure. You bring your clusters; we don't sell you infra. You get one place to spin up projects, deploy apps, and see logs and metrics, without the DevOps bottleneck. This page explains how it fits together.

Bring your own infrastructure

The value we provide: you keep using the infrastructure you already have (or the cloud provider you choose). You don't purchase clusters or hosting from us. You connect your clusters to Kaja, run a small agent where your workloads run, and from then on you develop and deploy from one dashboard—projects, apps, pause, resume, logs, metrics. Same workflow whether your infra is on-prem, in AWS, GCP, or local. Your infra. Your control. We simplify how you build and ship on it.

Runs on Kubernetes (no lock-in)

Under the hood, Kaja runs on Kubernetes — and we're upfront about it here in the docs, even though the dashboard hides the moving parts. Concretely:

• A cluster you connect can be any conformant Kubernetes cluster — managed cloud (EKS/GKE/AKS), on-prem, k3s, or a local k3d cluster.
• The agent is a small set of Kubernetes controllers (operators) installed into a namespace (default kaja) with Helm. It reconciles what you do in the dashboard into standard Kubernetes resources.
• A project maps to an isolated namespace; an app becomes a normal Deployment/Service, exposed through an ingress; secrets become Kubernetes Secrets; managed services are Helm releases.

Because it's plain Kubernetes underneath, there's no lock-in: your kubectl access keeps working, and if you ever stop using Kaja, your workloads keep running. Kaja's job is to remove the YAML and context-switching — not to take your cluster away.

What you get with Kaja

• One place to develop and deploy — Projects and apps in one dashboard. No switching contexts, no fragmented scripts. Spin up what you need for testing, staging, or production; self-serve.
• Your own project in seconds — Developers spin up and tear down their own isolated projects. Total autonomy, zero waiting on DevOps.
• Automated cost savings — Pause projects when you're done. Stop paying for idle resources; Kaja helps you keep cost under control.
• Enterprise-grade security — RBAC for teams and outbound-only agent connections. Run clusters with no public IP if you want—no open firewalls on the control path.

So from your point of view: simplified development and deployment on the infra you bring.

Two parts that work together

1. The app you use (the dashboard)
   The website you log into. You add the clusters you already have, create projects, deploy apps, view logs and metrics. The dashboard talks to each of your clusters through a small connector—the agent—so what you do in the app happens where your workloads run.

2. The agent (where your workloads run)
   When you connect a cluster, you run one install command on your cluster. That installs a small agent there. The agent carries out what you ask from the dashboard (create a project, deploy an app, pause, resume) and reports back so the dashboard stays in sync. One agent per cluster; each cluster registers with a one-time token (consumed once during install, then invalid) and then uses a refresh token for ongoing connection. Your infra stays yours; Kaja just makes it easier to use.

Where we're headed

We're focused on making projects, deployments, and observability faster on the clusters you already run. For enterprises that want the platform on their own infrastructure, we offer a self-hosted solution—see Self-hosted & enterprise.

Summary

Next steps

• Quick start — Connect your first cluster and deploy from the dashboard.
• Projects — Spin up your own project and pause when you're done. Apps — App features and managed services in depth.
• Connecting clusters — Add more of your clusters the same way.