Skip to content

JamesCallaghan/ai-ref-arch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ai-ref-arch

Prerequisites

To stand up the infrastructure for this demo, you need gcloud CLI access to a GCP project:

gcloud auth application-default login.

Passwords for several services will need to be set up. For simplicity, we have used the same password across all services. You will need to set the DEMO_PASSWORD environment variable to something of your choice:

export DEMO_PASSWORD=<your password here>

This demo uses the following client tools:

  • terraform
  • kubectl
  • helm
  • jq
  • envsubst
  • gcloud CLI
  • curl

Run the demo

Spin up a GKE cluster with Contour for Ingress, cert-manager to provide self-signed certificates for demo purposes, keycloak, the Helix ML control plane, and one Helix runner. This will take several minutes:

make all-up

To avoid the need to create DNS records just for our demo, we will use nip.io to access Helix from outside the cluster. Find the Contour LoadBalancer IP with the following command:

kubectl describe svc contour-envoy --namespace projectcontour | grep Ingress | awk '{print $3}'

The Helix UI will be available at https://<IP from the above command>.nip.io

You will be able to log in with the email xyz@example.com and the DEMO_PASSWORD you chose at the start.

Please note that the runner may take several minutes to be ready.

Testing

make test

Note that the test may fail the first time. Inspecting the runner logs may show Pulling model llama3:instruct. Once this is complete, the test should pass.

Teardown

make delete-all

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published