databricks as an api

[vaidhaicha]

Hi ,

In my databricks notebook , I would like to do some processing like calling an openAI model and generating the resonse. Now frontend(react)  will call my notebook and my notebook should be executed and generate some response .Now this response I would like to send it back to the frontend.

So what are the steps do we need to follow so that our azuredatabricks can be exposed as an endpoint

[Kaniz_Fatma]

Hi @vaidhaicha, Azure Databricks provides model-serving endpoints for deploying custom models. You can use the following methods to create endpoints:

• Serving UI:
  • In the Databricks workspace, click on “Serving” in the sidebar.
  • Create a serving endpoint by specifying the model, version, compute type, and other relevant settings.
• REST API:
  • Use the REST API to create endpoints programmatically. Refer to the API documentation for configuration parameters.
• MLflow Deployments SDK:
  • Install the MLflow Deployment client and create endpoints using the SDK.
  • Example: import mlflow.deployments; client = mlflow.deployments.get_deploy_client("databricks")