🔍🤖 RAG-Terraform-Vertex

---

Deploy a modern Retrieval-Augmented Generation (RAG) stack on Google Cloud using Terraform, Vertex AI, Discovery Engine, and GCS.

---

🚦 Deployment Options

You can deploy this RAG solution on Google Cloud in two ways:

• Option 1: Terraform/CLI (Recommended for automation)
• Option 2: Google Cloud Console (No CLI, UI only)

---

Option 1: Terraform/CLI Way

📚 What is RAG?

Retrieval-Augmented Generation (RAG) is an AI technique that enhances Large Language Models (LLMs) by combining information retrieval with generative capabilities. Instead of relying solely on model parameters, RAG retrieves relevant external documents from a knowledge base or data store, resulting in more accurate, contextual, and up-to-date responses.

---

🚦 Prerequisites

• 🏢 Google Cloud project with billing enabled
• 👤 IAM Role: Project Owner or Editor
• 🧑‍💻 Basic knowledge: Google Cloud, Terraform, Generative AI, LLMs, Embeddings

---

⚙️ Deployment Steps

1️⃣ Open Google Cloud Shell or Local Terminal

You can run the following commands in Google Cloud Shell or your local terminal (with gcloud CLI configured).

---

2️⃣ Clone the Repository

git clone https://github.com/anudishu/RAG-Terraform-Vertex.git
cd RAG-Terraform-Vertex

---

3️⃣ Enable Required Google Cloud APIs

gcloud services enable \
    compute.googleapis.com \
    aiplatform.googleapis.com \
    storage.googleapis.com \
    discoveryengine.googleapis.com

Verify APIs:

• Go to Google Cloud Console → Search for "AI Application"
• Navigate to the AI Applications page
• Click Continue and activate the API if prompted
• Confirm all required APIs are listed and active

---

4️⃣ Set Environment Variables

Replace placeholders with your actual values and execute it one-by-one

export PROJECT_ID="your-gcp-project-id" # get your project id from google cloud console
export REGION="us-central1"   # e.g. us-central1
gcloud config set project $PROJECT_ID

---

5️⃣ Authenticate with Google Cloud

Authenticate your environment to allow Terraform and Python scripts to access Google Cloud services:

gcloud auth application-default login

Follow the authentication flow:

1. Copy the provided link from the terminal
2. Open it in your browser and login with your GCP account
3. Copy the authorization code and paste it back in the terminal

---

6️⃣ Initialize and Apply Terraform Configuration

Before you deploy infrastructure:

Open the terraform.tfvars file and update the project_id variable with your actual Google Cloud project ID.

project_id = "your-gcp-project-id"

Save the file after making the change.

This will provision all required infrastructure on Google Cloud:

• Discovery Engine Data Store
• Discovery Engine Search Engine

terraform init
terraform plan
terraform apply -auto-approve

After successful deployment, note the Terraform outputs:

test_data_store_id = "demo_store_id"
test_engine_id = "demo_engine_app"

---

📥 Load Data into RAG Data Store

🔧 1. Update loaddata.py

Open loaddata.py and set the correct project ID at the end of the file:

PROJECT_ID = "your-gcp-project-id"

▶️ 2. Run the Data Loader

python loaddata.py

This script will upload public documents (gs://cloud-samples-data/gen-app-builder/search/alphabet-investor-pdfs) into your configured Vertex AI Search data store.

---

🧪 Test the RAG Application

Once your data is indexed:

• Send test queries using the deployed app interface (if applicable)
• Or trigger inference via Vertex AI API
• Ensure that answers reflect content from the uploaded documents

🖥️ Run a Query from Terminal

Open query.py, set the correct project ID at the end of the file, then run:

python query.py

💡 Sample Test Queries

QUERY = "who is ceo of google? what is total revenue of google?"

# More examples:
QUERY = "What were Google Cloud earnings in 2024?"
QUERY = "Google Cloud financial results and revenue growth"
QUERY = "What is Google's market capitalization and stock performance?"
QUERY = "Google advertising revenue breakdown by quarter"
QUERY = "Alphabet Inc financial highlights and key metrics"
QUERY = "Google Cloud vs AWS market share comparison"
QUERY = "What are Google's main business segments and revenue sources?"
QUERY = "Google's investment in AI and machine learning initiatives"
QUERY = "YouTube revenue and user engagement statistics"
QUERY = "Google's data center locations and infrastructure investments"

---

🧹 Clean Up Resources

To avoid incurring charges, destroy the provisioned infrastructure when no longer needed:

terraform destroy -auto-approve

---

📌 Additional Notes

• Ensure your IAM permissions allow you to create and manage required resources.
• For troubleshooting, check the Google Cloud documentation or open an issue in this repo.

---

Happy Building! 🚀

---

Option 2: Google Cloud Console Way

No CLI or Terraform required! Use the Google Cloud Console UI for a fully guided, point-and-click deployment.

🖱️ Deploy RAG-Terraform-Vertex Using Google Cloud Console (No Terraform/CLI)

This guide walks you through deploying a Retrieval-Augmented Generation (RAG) application using only the Google Cloud Console (web UI). No command line or Terraform is required. Follow these steps to complete the hands-on lab:

---

1️⃣ Create or Select a Google Cloud Project

1. Go to the Google Cloud Console.
2. In the top navigation bar, click the project dropdown and select an existing project or click New Project to create one.
3. Ensure billing is enabled for your project. (Enable billing)

---

2️⃣ Enable Required APIs

1. In the left menu, go to APIs & Services > Library.
2. Search for and enable each of the following APIs:
   • Vertex AI API (aiplatform.googleapis.com)
   • Discovery Engine API (discoveryengine.googleapis.com)
   • Cloud Storage API (storage.googleapis.com)
   • Compute Engine API (compute.googleapis.com)

---

3️⃣ Set Up Vertex AI Search (Discovery Engine)

1. In the Google Cloud Console, search for AI Application.
2. Under Datastores, select Create Data stores.
3. Select Data Source as Cloud Storage.
4. Leave all settings as default. For the GCS URL, enter:
   • cloud-samples-data/gen-app-builder/search/alphabet-investor-pdfs
5. Fill in the required details:
   • Datastore Name: Enter a name for your data store.
6. Click Create and wait for the data store to be provisioned.
7. Note your Data Store ID for later use.

---

4️⃣ Create a Search Engine

1. In AI Application, go to the right-side menu and click App > Create App.
2. Select Custom Search.
3. Enter an App Name.
4. Enter a Company Name (you can use any value).
5. Click Continue.
6. Attach the previously created datastore.
7. Done!

---

Test Your RAG Application

---

Clean Up Resources

---

Note: All steps above use only the Google Cloud Console UI. For advanced automation, see the Terraform-based instructions above.