How to integrate external ML model outputs into GraphHopper backend?

Directions API Cluster API
1

Hi,

I am working on a project where we use an external Machine Learning model to process user data. This ML model provides specific values (such as predicted clusters or weights) that need to be fed into the GraphHopper backend.

Currently, I am having trouble identifying exactly which part of the GraphHopper backend source code I should modify to inject these processed values.

Specifically, I would like to know:

  1. Integration Point: Where is the best place in the backend logic to intercept the request and inject the ML model’s output before the routing or clustering begins?

  2. File References: Which classes or files are typically responsible for handling external data inputs in the graphhopper-web or core modules?

  3. Best Practice: Should I handle this at the API/Controller level, or is there a more idiomatic way to customize the Cluster API or Matrix API logic to account for these external values?

I’m a bit lost on where to start modifying the code. Any guidance on which files to look into would be greatly appreciated!

Best regards,