Catalytic AI fine-tunes language models on client data that cannot be sent to a cloud API, and builds the on-premise machines those models run on. Work stays inside the client's building: no hosted endpoint, no third-party API in the request path, and the client keeps the weights, the training pipeline and the evaluation harness.
Hourly rate: From $100/hr