A production-grade RAG (Retrieval-Augmented Generation) system for querying Ethereum blockchain data using natural language, built entirely with BigQuery ML and Google APIs.

📄 Read the Technical Write-up - A comprehensive guide on building RAG systems using only SQL and Google APIs.

This guide uses the Application Default Credentials (ADC) method.

1. Set up a Google Cloud project

   • Create a Google Cloud project
   • Enable the Vertex AI API

2. Install the gcloud CLI

   • macOS: brew install --cask google-cloud-sdk
   • Windows: gcloud CLI installer

1. Environment variables
   Create a .env file with your project and location settings.

   • Recommended: use US as the location since public datasets are in the US region (this avoids cross-regional issues).
   • Use .env.example as a template.

2. Authenticate with Google Cloud
   Run the following commands in your terminal:

   # Load environment variables
   set -a; source .env; set +a;
   
   # Authenticate
   gcloud auth application-default login
   
   # Set project
   gcloud config set project "$GOOGLE_CLOUD_PROJECT"

3. Create a BigQuery connection

   bq mk --connection \
   --project_id="$GOOGLE_CLOUD_PROJECT" \
   --location="$GOOGLE_CLOUD_LOCATION" \
   --connection_type=CLOUD_RESOURCE \
   bq-llm-connection || true
   
   echo '---'
   
   bq show --connection \
   --project_id="$GOOGLE_CLOUD_PROJECT" \
   --location="$GOOGLE_CLOUD_LOCATION" \
   bq-llm-connection || true

   More on BigQuery connections here

If you encounter the following error while running BigQuery ML models:

The [email protected] does not have the permission to access or use the endpoint. Please grant the Vertex AI user role to the [email protected] following https://cloud.google.com/bigquery/docs/generate-text-tutorial#grant-permissions. If issue persists, contact [email protected] for help.

Grant the Vertex AI User role to the service account:

• Follow this guide.

.
├── bigq_ai_app/
│   ├── core/
│   ├── utils/
│   └── ui.py
├── docs/
├── notebooks/
├── tests/
├── .env.example
├── .gitignore
├── app.py
├── pyproject.toml
└── README.md

1. Create a virtual environment:

   uv venv

2. Activate the environment:

   source .venv/bin/activate
   # On Windows: .venv\Scripts\activate

3. Install dependencies from your pyproject.toml:

   uv pip install -e .

   (The -e . installs your project in "editable" mode, which is great for development).

4. Run the Gradio app:

   python app.py

   • Open http://127.0.0.1:7865
   • To expose a public link quickly: set share=True in launch()
   • To bind for containers/VMs: use server_name="0.0.0.0