vertex ai gemini integrator

I have generated a production-ready Python script using the vertexai SDK. This implementation uses Application Default Credentials for secure authentication and incorporates the tenacity library to handle potential API rate limits with exponential backoff.

Vertex AI Structured Output Implementation

import json
import vertexai
from vertexai.generative_models import GenerativeModel, GenerationConfig
from tenacity import retry, stop_after_attempt, wait_exponential, retry_if_exception_type

# Initialize Vertex AI with your project and location
vertexai.init(project="your-project-id", location="us-central 1")

@retry(
    stop=stop_after_attempt(5),
    wait=wait_exponential(multiplier=1, min=4, max=10),
    retry=retry_if_exception_type(Exception)
)
def extract_action_items(transcript: str):
    """
    Extracts action items from a transcript and enforces JSON output.
    Uses ADC for authentication - ensure GOOGLE_APPLICATION_CREDENTIALS is set.
    """
    model = GenerativeModel("gemini-1.5-flash")
    
    prompt = f"Extract all action items from the following transcript: {transcript}"
    
    # Strict JSON enforcement via GenerationConfig
    response = model.generate_content(
        prompt,
        generation_config=GenerationConfig(
            response_mime_type="application/json"
        )
    )
    
    try:
        return json.loads(response.text)
    except json.JSONDecodeError:
        return {"error": "Failed to parse model output as JSON"}

# Example call
transcript_data = "Meeting notes: Sarah will fix the login bug by Friday. Mike needs to update the docs."
actions = extract_action_items(transcript_data)
print(json.dumps(actions, indent=2))

Next steps

1. Ensure the Vertex AI API is enabled in your Google Cloud Console.
2. Run gcloud auth application-default login to set up your local environment.
3. Install dependencies: pip install google-cloud-aiplatform tenacity.