Lightweight server-based NLU using Rasa?

I need to train a large model on a desktop, then move access the model to the web to access it in real time on a Linux (Debian) AWS server.

All I want to do is extract entities and intents from the model. Is there a a lightweight version that just queries the trained model available, or do I need to install the full Rasa distribution? Is there a server built in, or do I need to write one?