ML engineers often create their own models using open-source libraries, like TensorFlow running on GCP.

The value of these models can be huge, if you build and train them correctly or minimal if they are not done well.