[Enhancement] Support Vertex AI, Bedrock Models

[psriramsnc]

🚀 Add Support for Vertex AI, Bedrock Models

✨ Summary

This PR expands and upgrades SyGra’s support for LiteLLM-based models by adding multiple new backend integrations (Ollama, Azure, Vertex AI, Triton, Bedrock), refactoring the LiteLLM model abstractions into a cleaner, more modular model factory, and adding documentation + unit tests for the new model paths. This broadens SyGra’s flexibility and makes it easier to hook into various LLM-serving providers.

✅ Features implemented

• Support for new LiteLLM providers
  • Ollama
  • Azure
  • Triton
  • Vertex AI
  • Bedrock
• Refactored LiteLLM model factory / abstraction layer
  • Consolidated model instantiation logic so all providers reuse shared code paths
  • Cleaner, more maintainable, and easier to extend for future providers.
• Documentation updates
  • Added/updated docs to explain how to configure and use the newly supported LiteLLM providers.
• Test coverage
  • Added unit tests for Vertex AI and other backend integrations to ensure correctness and stability

⚡ Performance impact (if any)

• Minimal or none — since this is primarily an architectural/refactor and integration PR.
• Cleaner abstractions may slightly reduce overhead; logging and config handling should remain efficient.
• No regressions in runtime performance expected.

🧪 How to Test the feature

Steps for reviewers to verify the new backend support:

1. Update models.yaml (or equivalent config) to point to one of the newly supported providers (e.g. Ollama or Vertex AI).
2. Run a sample SyGra workflow (or minimal test case) that invokes an LLM node using the configured model.
3. Verify that:
   • The model instantiates correctly via the LiteLLM factory.
   • The inference request completes successfully without errors.
   • The output matches expectations for that backend.
   • Logging and environment configuration (URLs, tokens, headers) works correctly.
4. Run the unit tests (including those added for Vertex AI) and confirm they pass.

📸 Screenshots (if applicable)

N/A

📝 Checklist

• Lint fixes and unit testing done
• End-to-end task testing (e.g. full SyGra workflow)
• Documentation updated

🔎 Notes

• Backward compatibility remains intact — existing workflows should continue working unchanged.

[amitsnow]

LGTM 🚀