I need to configure the multi-region Kubernetes deployment. My services use IBM Watson. But Watson does not provide a global instance. It based on the region. Am I to use two different Watson for two regions?
Depending on why you want to deploy a multi-region app and what type of IBM Watson service you want to integrate, there are different options available. There is an IBM Cloud solution tutorial with strategies for resilient applications which might be a good introduction and with related links.
If it is for resiliency, you would need to check what SLAs and deployment model the service in question offers. Depending on the IBM Watson service the APIs are stateless or require to open a session. Thus, the application design needs to take that into account.
If it is for performance for a global audience (app users), you might need to look into how to split traffic, cache answers, or deploy services and app instances closer to the user.
Without any details from your side it is a pretty broad question and hard to answer.