migrate-from-model-serving

Migrate MLflow agents to Databricks Apps by transforming class-based predict methods into decorated functions. This Skill automates converting ResponsesAgents and configures Databricks Asset Bundles for deployment on the new platform.

Converting MLflow predict methods to Databricks Apps decorators involves transforming predict and predict_stream methods into invoke and stream decorated functions. The Skill handles this code transformation for both synchronous and asynchronous execution.

Yes, you can migrate MLflow agents to an asynchronous pattern when moving to Databricks Apps. The Skill optionally converts your code to support asynchronous operations for improved concurrency during deployment.

Migrating to Databricks Apps handles downloading and reapplying original model artifacts and code dependencies. The Skill ensures your GenAI agent retains its required components throughout the deployment modernization process.

Moving GenAI agents from Model Serving to Databricks Apps provides a more flexible deployment environment with an integrated UI, easier deployment processes, and enhanced observability features for monitoring agent performance.

Transforming MLflow ResponsesAgents for Databricks Apps focuses specifically on converting class-based predict methods to decorated functions. Migration scope is limited to the Databricks ecosystem and requires configuring Databricks Asset Bundles for deployment.