An end to end ML pipeline

To run

1. Because we are using two distinct ports (one for the site layout, one for API calls), we must tell k3d to map both external ports when creating the cluster. Build both images locally:

docker build -t my-flask-backend:latest ./backend
docker build -t my-html-frontend:latest ./frontend

2. Create the K3d Cluster exposing two ports: Port 8080 maps to the frontend service (Nginx). Port 8081 maps to the backend service (Flask).

k3d cluster create mycluster -p "8080:80@loadbalancer" -p "8081:5000@loadbalancer"

3. Import images directly into k3d using imagePullPolicy:

k3d image import my-flask-backend:latest my-html-frontend:latest -c mycluster

4. Deploy:

kubectl apply -f k8s/

5. Navigate to http://localhost:8080 for frontend and port 8081 for querying backend.