Skip to content

v0.4.0

Choose a tag to compare

@geronimo-deploy-cloud geronimo-deploy-cloud released this 15 Feb 23:57

Geronimo v0.4.0 Release Notes

This release focuses on consolidation and simplification. We've significantly reduced the API surface area, removed legacy components, and unified the deployment workflow around Geronimo Deploy Cloud (GDC).

⚠️ Breaking Changes

  • Renamed Cloud Backend: The cloud backend has been renamed to deploy_cloud to clearly distinguish the Geronimo Deploy Cloud platform from generic cloud providers.
    • Action Required: Update your geronimo.yaml deployment target from cloud to deploy_cloud.
  • CLI Cleanup: Removed geronimo import and geronimo monitor commands. Monitoring is now natively integrated via the SDK and GDC.
  • Data Source Refactoring: The src/geronimo/data directory has been renamed to src/geronimo/data_sources to better reflect its purpose.

🚀 New Features

Secret Management

  • geronimo keys sync: A new command to securely synchronize local keys/secrets to Geronimo Deploy Cloud, simplifying environment variable management for deployed models.

Enhanced SDK

  • Multi-Source Data: Added support for defining multiple data sources and auto-joining them in the SDK.
  • Improved Model Templates: model.py templates now include comprehensive scaffolds for save, load, train, and predict methods, reducing boilerplate.
  • Agentic Workflows: Realtime projects now leverage fastmcp for high-performance Model Context Protocol (MCP) server integration.

Configuration

  • Artifact Store Setup: Added a new configuration command to streamline the setup of the Artifact Store.

🛠 Improvements

  • Jinja2 Templating: Migrated Project, Docker, Pipeline, and Terraform generators to use Jinja2 templates. This ensures more robust code generation and easier extensibility.
  • Unified Documentation: Centralized module definitions and improved the API documentation generation process for better consistency.
  • Refactored Codebase: Extensive cleanup and consolidation of internal modules to improve maintainability and performance.