I am a beginner and I have a Flask application(Computer Vision use case) and am required to deploy it on a VM scale set(N-series GPU instances) since I’d require the auto-scaling functionality(we don’t know how many clients would require this product and hence can’t estimate the customer pool).

This’ll be my first time deploying an application on a VM directly(instead of using the App Service) and I don’t know how to proceed with it. Alongside the VMSS service, I also created the Load balancer and VNet service but am not knowledgeable about how to configure . I can easily deploy the Flask app on a single instance and access it using the instance’s public IP. However, I wish to make it accessible via the load balancer’s IP so that the service can pass the traffic to the required instance when it auto-scales. Please let me know how this can be done.

I’m also required to add startup scripts to each VM when it gets created. Is it possible via ARM/Terraform?