RAG-based Flight Query Bot

Project Overview

The RAG-based Flight Query Bot is a Streamlit-powered web application that allows users to query flight information using natural language (e.g., "Show me flights from New York to London"). It uses Retrieval-Augmented Generation (RAG) with an Ollama language model to extract entities from queries and retrieve flight data from a mock database. The project is deployed on a local Kubernetes cluster using Minikube and includes a CI/CD pipeline with GitHub Actions for automated testing.

Architecture

  RAG-based-Flight-Query-Bot/
  ├── .github/
  │   └── workflows/
  │       └── test.yml
  ├── tests/
  │   └── test_mock_database.py
  ├── mock_database.py
  ├── query_handler.py
  ├── ollama_api.py
  ├── app.py (optional)
  ├── requirements.txt
  └── README.md

• Frontend: Streamlit UI (app.py) for user interaction.
• Backend:
  • query_handler.py: Processes queries and extracts entities using Ollama.
  • ollama_api.py: Integrates with the Ollama LLM for natural language responses.
  • mock_database.py: Provides mock flight data and search functionality.
• Deployment: Kubernetes on Minikube with two services: flight-assistant-service (Streamlit) and ollama-service (Ollama server).
• CI/CD: GitHub Actions runs unit tests on every push or pull request.

---

Setup Instructions

Prerequisites

• Windows, macOS, or Linux
• Python 3.10 or 3.11
• Docker Desktop (for building images and running Minikube)
• Minikube (for local Kubernetes)
• kubectl (Kubernetes CLI)
• Git (for cloning the repository)

Step 1: Clone the Repository

Clone this project to your local machine:

git clone https://github.com/RobuRishabh/RAG-based-Flight-Query-Bot.git
cd RAG-based-Flight-Query-Bot

Step 2: Install Python Dependencies

Create a virtual environment and install required Python packages:

# Create a virtual environment (optional but recommended)
python -m venv flightquerybot

# Activate it (Windows)
flightquerybot\Scripts\activate

# Activate it (macOS/Linux)
source flightquerybot/bin/activate

# Install dependencies
pip install -r requirements.txt

Step 3: Set Up and Run the Local Ollama Server

The project uses a local Ollama server for language processing.

Install Ollama:

• Download and install Ollama from ollama.com.
• Follow the installation instructions for your OS.

Pull the Model:

ollama pull qwen2.5-coder:3b

This downloads the qwen2.5-coder:3b model (default for this project).

Run the Ollama Server:

ollama serve

Keep this terminal open. The server runs at http://localhost:11434.

Step 4: Set Up Minikube

Minikube creates a local Kubernetes cluster.

Install Minikube:

Follow the official Minikube installation guide for your OS. Example for Windows (using PowerShell as admin):

Start Minikube:

minikube start

Verify Minikube:

minikube status

Ensure host, kubelet, and apiserver are Running.

---

Deployment Steps

Step 1: Build the Docker Image

Build the flight-assistant Docker image using Minikube’s Docker daemon:

# Set Minikube's Docker environment
eval $(minikube docker-env)

# Build the image
docker build -t flight-assistant:latest .

Step 2: Apply Kubernetes YAML Files

Deploy the application and Ollama server to Minikube:

# Deploy Ollama server
kubectl apply -f ollama-deployment.yaml
kubectl apply -f ollama-service.yaml

# Deploy Flight Assistant app
kubectl apply -f deployment.yaml
kubectl apply -f service.yaml

Step 3: Expose and Access the Services

Start Minikube Tunnel:

minikube tunnel

Keep this running to assign an external IP.

Get the Service URL:

minikube service flight-assistant-service --url

Example output: http://192.168.49.2:8501.

Access the App:

Open the URL in your browser to interact with the chatbot.

---

Usage

Interacting with the Chatbot UI

1. Open the Streamlit app in your browser.
2. Enter a query in the text box and press Enter.
3. The bot processes your query and displays flight details or a "no flights found" message.

Example Queries

• "Show me flight NY100" → Shows details for flight NY100.
• "What are the flights from New York to London?" → Lists matching flights.
• "Flights from Chicago" → Shows flights departing from Chicago.
• "Flight XYZ999" → Returns "No flights found" (invalid flight).

---

CI/CD Explanation

GitHub Actions Configuration

This project uses GitHub Actions for Continuous Integration (CI) to ensure code quality. The workflow is defined in .github/workflows/test.yml.

Workflow Details

• Trigger: Runs on every push or pull_request to the main branch.
• Environment: Uses an ubuntu-latest runner with Python 3.10.
• Steps:
  • Checkout code:

    - uses: actions/checkout@v4

  • Set up Python:

    - uses: actions/setup-python@v5
      with:
        python-version: '3.10'

  • Install dependencies:

    - run: |
        python -m pip install --upgrade pip
        pip install -r requirements.txt

  • Run tests:

    - run: |
        python -m unittest discover -s tests -p "test_*.py"

---

Running Tests Locally

To test before pushing:

# Activate virtual environment
flightquerybot\Scripts\activate  # Windows
source flightquerybot/bin/activate  # macOS/Linux

# Run all tests
python -m unittest discover -s tests -p "test_*.py"

Expected output: Ran X tests in Y.YYYs OK.

---

Documentation Notes

Stopping Minikube

minikube stop

Cleanup (Reset Everything)

minikube delete

Troubleshooting

If the UI doesn’t load, check pod logs:

kubectl logs -l app=flight-assistant
kubectl logs -l app=ollama

Ensure minikube tunnel is running for external access.

---