In this step, you will need to deploy the CD pipeline that will build the Jetson Operating System image, including Microshift as well as the five key components of the train:
This pipeline will also trigger an update of the fleet corresponding to your Edge device. Each participant has a virtual machine “at the Edge”, connected to Red Hat Edge Manager, the Edge device fleet management system.
To help you, a Helm Chart is present in the application mono repo (tekton-pipelines folder).
This Helm chart contains a Tekton pipeline that builds the Jetson Operating System image and triggers the update of your Edge device.
You will deploy the tekton pipeline from your OpenShift DevSpaces environment (it will be easier).
Connect to the Red Hat Edge Manager console with your username and password.
You should see 40 Edge devices and 40 fleets, one device and one fleet per participant.
Navigate to Fleets > your fleet > Actions > Edit fleet configuration

In the Device template section, under Host configuration (files), click on Add configuration.
Fill in the form with the following values:
train-config/etc/default/bootstrap-microshiftrootrootkafkaBroker.username=train
kafkaBroker.password=R3dH4t1!
kafkaBroker.bootstrapNode.hostname=ae8f26017d0a9453ba7a403b7794c299-862939278.eu-central-1.elb.amazonaws.com
Your configuration should look like this:

Click on Next > Next and Save.
On your fleet, click on the X/Y under Up-to-date/devices to access your Edge device list.

Click on your single Edge device.
Click on Terminal. Wait for the Edge Device to contact the Edge Manager server (it can take up to a minute!).
Once the terminal is connected, execute the following command:
cat /etc/default/bootstrap-microshift
And confirm that your configuration file has been deployed.

In this step we will start a tekton pipeline that will build the Edge device Operating System image and trigger its update via Red Hat Edge Manager.
To do this, open a terminal in OpenShift DevSpaces.
From the terminal, discover the projects you have access to.
oc get projects
You should see three OpenShift projects:
$USERID-devspaces)$USERID-test)$USERID)Get the test project name in an environment variable.
TEST_NS=$(oc get projects -o name -l env=test | cut -d / -f 2 | head -n 1)
echo "Using namespace $TEST_NS"
In your DevSpaces workspace, edit the file tekton-pipelines/values.cd.yaml and replace the occurrences of userXY with your username.
There are normally two occurrences to replace.
Create the PipelineRun object in your OpenShift test project.
helm template pipelines /projects/summitconnect2025-app/tekton-pipelines --set namespace="$TEST_NS" --values /projects/summitconnect2025-app/tekton-pipelines/values.cd.yaml | oc create -f -
The warning message “WARNING: Kubernetes configuration file is group-readable. This is insecure." can be ignored.
Open the OpenShift console and navigate to Administrator > Pipelines > Pipelines > PipelineRuns.
Normally, the pipeline should start immediately.

The pipeline takes about 12 minutes to complete. Once the pipeline is finished, you should see that the fleet in Red Hat Edge Manager has been updated with the image generated by the pipeline.

On your fleet, click on the X/Y under Up-to-date/devices to access your Edge device list.

Click on your single Edge device.
You should see that your Edge device is in “out-of-date” state.

Wait a few minutes while it downloads its update, applies it and reboots.
Navigate to Fleets > your fleet
On your fleet, click on the X/Y under Up-to-date/devices to access your Edge device list.

Click on your single Edge device.
Click on Terminal. Wait for the Edge Device to contact the Edge Manager server (it can take up to a minute!).
Once the terminal is connected, execute the following command:
export KUBECONFIG=/var/lib/microshift/resources/kubeadmin/kubeconfig
oc -n train get pods -w
You should see the containers constituting the train autopilot, including the intelligent-train which contains the AI model.
Congratulations, you just deployed an AI model to the Edge! 🎉