Demo Readydemo-0023
Kamiwaza — AI Orchestration Platform on HA Kubernetes
Kamiwaza AI orchestration on a four-node HA Kubernetes cluster, one node per hypervisor, in the demo landing zones.
computegpuaikuberneteskamiwazain-progress
Required Resources
demosvr01demosvr02demosvr03demosvr04coresan01
What a customer gets out of it
- Show an AI orchestration platform running on customer-owned infrastructure, not a hosted API
- Demonstrate a four-node HA Kubernetes cluster surviving the loss of a control-plane node
- Let a customer deploy and serve their own models against their own data, inside the lab
Why it matters
Most AI conversations stall on where the data goes. Kamiwaza runs the orchestration layer on hardware the customer controls, so models are served next to the data rather than shipped to a vendor — which turns a governance objection into an architecture discussion. Built on a genuinely highly-available Kubernetes cluster across four hypervisors, it also answers the second question that follows: whether the platform underneath is production-shaped or a single-node demo.
Lab overview
No walkthrough is recorded for this demo yet, so this is a 21 sec tour of the lab and how to use it.
Trouble playing? Download the file.
**Kamiwaza** is an AI orchestration platform, deployed here on infrastructure the customer controls rather than consumed as a hosted API.
Each landing zone builds **one highly-available Kubernetes cluster spread across all four demo hypervisors** — three control-plane nodes behind a keepalived API VIP, plus a worker — so a control-plane node can be lost mid-session without taking the platform down. Kamiwaza itself is reached through a MetalLB service VIP.
Use it to show model serving and orchestration running next to the customer's own data, and to answer the question that always follows: whether the platform underneath is production-shaped or a single-node demo.