Run your model in Melbourne or in your own data centre
Melbourne needs no hardware from you, and your own data centre, connected or air-gapped, keeps the traffic on your network.
- Hardware
- We supply it, in our Melbourne facility.You supply anything from one multi-GPU node to a full rack, or we build a cluster to our reference design and hand it over.
- Your data
- It stays in your instance in Australia, and no other customer’s model trains on it.It stays on your network, with no route from us to your traffic.
- Updates
- A new version goes live only when you approve it.Signed artefacts arrive offline, with release notes, evaluation results and a rollback.
Wherever it runs
You keep the same four controls in each option.
They hold in Melbourne, in your own rack and in a room with no network.
You hold the weights.
You hold them under the base model’s licence, and they leave with you.
You approve each version.
A new version goes live only after you approve it.
Your sign-in and your records.
Staff sign in through your identity provider, and history, metering and the audit trail stay in your instance.
Australian law, in the contract.
We’re Australian-owned with no foreign parent, and the contract puts your data under Australian law.
Rollout
Five steps from scoping to operation.
We run it as an infrastructure project, paced by your change board and your assessors.
- 01
Scope
We agree the task, its security classification and the hardware you already have.
- 02
Design
We draw up a reference architecture covering model sizes, node counts, network and identity.
- 03
Install
We install it with your platform team, through your change process.
- 04
Accredit
We supply the evidence pack and sit with your assessors.
- 05
Operate
We benchmark each update before it ships, release it on your schedule, and give you a named engineer to call.
Scoping