Self-Hosting VDF AI

Why some teams run VDF AI on their own infrastructure

VDF AI is available as a managed service for most teams. But for organizations with strict residency, sovereignty, or compliance constraints, self-hosting is a first-class deployment option — the same product, running inside your own environment.

You’d choose self-hosting when:

• Your data can’t leave your network — regulated industries, classified workloads, customer data with residency clauses.
• Your security model requires the AI stack to live behind your firewall.
• You’re running in an air-gapped or hybrid-cloud topology.
• You want full control over the inference path, the audit trail, and the operational stack.

The portal — your hub for everything self-hosted

Self-hosting starts at portal.vdf.ai. The portal is a secure access hub: you register your organization, get verified by our team, and receive signed, production-ready container packages you can deploy on your own infrastructure.

From the portal you can:

• Register your organization and provision an account.
• Browse the catalog of self-hostable packages.
• Generate time-limited download credentials.
• Track which packages your team has deployed and at what version.
• Request trial extensions, upgrades, and support.

The portal does not deploy for you — it gives you the artifacts and the path. The deploy itself happens in your environment, on your terms.

Who self-hosting is for

• Regulated industries (FSI, healthcare, defense)
• Public sector & government
• Data-sovereign teams
• Air-gapped operations
• Hybrid-cloud architects
• Platform / SRE teams
• Security & compliance leads

A typical buyer set:

• The compliance lead owns the “why” — residency, AI Act readiness, audit posture.
• The platform / SRE team owns the “how” — sizing, orchestration, observability, updates.
• The security team owns the gates — image provenance, network policies, secret handling.

What’s available today

VDF AI publishes self-hostable software as packages. Each package is a curated bundle of services and configuration designed to deploy together as one product.

The package catalog is the place to start. See Packages for what’s currently shippable and what’s planned.

What a deployment looks like, in five steps

1. Register at portal.vdf.ai.

   Provide your organization details. Verify your email. Wait for approval (typically one business day).

2. Pick a package.

   Browse the catalog inside the portal. Read the package's system requirements and confirm your target environment fits.

3. Generate credentials.

   One click in the portal mints time-limited credentials that let you pull the package's container images.

4. Deploy in your environment.

   Pull the images. Bring them up with Docker Compose, Kubernetes, or the orchestrator of your choice. Validate the services are healthy.

5. Hand off to your team.

   Configure user access on the deployed instance. Connect sources and integrations. Your users now have VDF AI inside your perimeter.

A typical first deployment lands in a few hours of operator time, most of it waiting for image pulls. See Getting started for the end-to-end walkthrough.

What you’ll need

You don’t need any pre-existing VDF AI relationship to start. You’ll need:

• A portal.vdf.ai account (free; created at sign-up).
• A target environment — a Linux host with Docker, or a Kubernetes cluster, or a hybrid setup.
• Outbound network access to pull container images (or a plan for air-gapped delivery).
• Someone with the access to run a deploy in your environment.

For the full list of compute, network, and storage requirements, see Infrastructure requirements.

Self-hosting vs. managed VDF AI

Both deploy the same product. The choice is mostly about who owns the operational surface.

Concern	Managed VDF AI	Self-hosted VDF AI
Where your data lives	VDF AI cloud	Your environment
Who runs the stack	VDF AI	Your platform team
Update cadence	Continuous	You pull when ready
Network model	Internet-accessible	Inside your perimeter
Compliance / residency	Standard contracts	Full control
Best for	Most teams	Regulated / sovereign / air-gapped

If you’re not sure which fits, start with managed — it gets your team productive in minutes. Move to self-hosted when a specific constraint requires it.

What to read next

• Getting started — the end-to-end walkthrough, from sign-up to first running deployment.
• The portal — how portal.vdf.ai works: accounts, approvals, credentials.
• Infrastructure requirements — compute, network, storage, and orchestrator details.
• Packages — the catalog of self-hostable software.
• Operations — updates, monitoring, scaling, and getting help once you’re running.
• FAQ — approvals, multi-user, air-gapped, licensing.