Kuldipgodase07/DataPulse

Model Context Protocol (MCP) Server on GCP

Unlock AI-Driven Insights with DataPulse on Google Cloud Platform

A FastAPI-based MCP server that enables AI agents to perform real-time data analytics on Google Cloud Platform, integrating with Vertex AI and BigQuery.

Architecture Diagram

Key Features

• Unified Query & Inference API: One server, two superpowers — run SQL analytics on BigQuery or get instant predictions from Vertex AI models.
• Blazing Fast: Asynchronous FastAPI backend for high-performance, low-latency requests.
• Cloud-Native Integration: Native support for Google Cloud authentication and resource management.
• Developer Friendly: Clean, well-documented REST endpoints, ready for integration with any modern tech stack.
• Secure by Design: Leverages GCP IAM for secure data access and model invocation.

Project Structure

• main.py – Main FastAPI server code
• requirements.txt – Python dependencies

Setup

1. Create and Activate a Virtual Environment:

   python3 -m venv venv
   source venv/bin/activate
   

2. Install Dependencies:

   pip install -r requirements.txt
   

3. Configure GCP Credentials:

   • Set your GOOGLE_APPLICATION_CREDENTIALS environment variable to point to your service account JSON file:

     export GOOGLE_APPLICATION_CREDENTIALS="path/to/your/service-account.json"
     export GCP_PROJECT="your-gcp-project"
     export GCP_LOCATION="us-central1"
     export BQ_DATASET="your_bigquery_dataset"
     

4. Run the Server:

   uvicorn main:app --reload
   

Usage

Query BigQuery

POST /query

Body:

{
  "sql": "SELECT * FROM your_dataset.your_table LIMIT 10"
}