Before You Start

What needs to be decided before a BullSequana AI deployment starts

Agentic Friendly

This page is the starting point for preparing a BullSequana AI deployment.

Its purpose is not to replace the detailed playbooks, but to clarify the decisions that need to be made before platform bootstrap begins.

What Must Be Clear Up Front

  • target deployment model
  • cluster and storage foundations
  • DNS and certificate ownership
  • registry and Git delivery path
  • identity and access assumptions
  • enabled platform scope

Cluster Foundations

Before bootstrap, the target Kubernetes environment should already have:

  • working cluster access
  • a chosen ingress class
  • valid ReadWriteOnce and ReadWriteMany storage classes
  • GPU capacity if inference services are part of scope
  • a clear node placement strategy if workloads need affinity or isolation

DNS and Certificates

The platform expects a base domain and a certificate strategy.

That means deciding:

  • which DNS zone will host the platform endpoints
  • who manages DNS updates
  • whether certificates are provided directly or issued through DNS automation
  • whether the target environment needs cloud-specific DNS integration

Registry and Git Delivery Path

Deployment also depends on two external delivery systems:

  • a container registry for platform images and Helm artifacts
  • a Git repository that Argo CD can reconcile from

In some environments those services already exist. In others, they need to be prepared as part of the delivery process.

Identity And Access

Key identity assumptions should be settled early:

  • standalone platform identity or federation with an existing IdP
  • expected access model for operators, developers, and end users
  • client and secret provisioning for platform-facing services

Platform Scope

Decide what is actually in scope for the first rollout:

  • Runtime only
  • Runtime + CoreAI
  • Runtime + CoreAI + ProAI
  • additional use cases or extensions

That decision affects enabled components, secrets, storage needs, and rollout complexity.

  1. Choose a Deployment Model
  2. Prerequisites
  3. Environment Guides
  4. Configuration Model
  5. Deployment Sequence
  6. Artifact Delivery with Harbor
  7. Troubleshooting

Once these decisions are clear, the deployment work becomes much more predictable.

Choose a Deployment Model

How BullSequana AI can be deployed across hardware, sovereign clouds, and existing infrastructure

Prerequisites

Which foundations, delivery services, and shared platform inputs must be ready before BullSequana AI deployment begins

On this page

What Must Be Clear Up FrontCluster FoundationsDNS and CertificatesRegistry and Git Delivery PathIdentity And AccessPlatform ScopeRecommended Reading Order