It looks like MLFlow doesn't natively support any authentication schemes and the recommendation is to use a reverse proxy such as Nginx to only forward requests to MLFlow if a valid authentication cookie is provided, redirecting any requests without a cookie or an invalid cookie back to the Identity Provider.
Azure AD Application Proxy
You can configure Azure AD Application Proxy to add authentication, although you will also need to protect MLFlow from unauthorised access by putting it in a virtual network or creating rules to only accept connections from Azure AD Application Proxy.
Used managed ML Flow in Azure
There is also the potential of using MLFlow that is part of Azure Databricks, which probably also includes authentication.
There are complete instructions on setting this up on the Nginx for a generic application. In summary, however, you will need to:
- configure Nginx as a reverse proxy
- prevent access to MLFlow from anything other than Nginx
- create an application in Azure Active Directory for Nginx
- configure Nginx to look for the JWT token and validate it
You would also be able to use other products such as Google's Identity Aware Proxy and ForgeRock's Identity Gateway.