Flask-ML-Deployment

Objective

This project aims to automate a distributed pipeline for AI applications by:

• Deploying AI models using Flask.
• Securing the application.
• Containerizing the application with Docker.
• Implementing multiple model endpoints (Object Recognition, Regression, and Text Generation).
• Automating deployment processes.

---

Implementation Details

Step 1: Deploying a VGG16 Object Recognition Model

• Endpoint: /predict
• Description:
  • Accepts an image file.
  • Processes it using the pre-trained VGG16 model.
  • Returns the object label and confidence score.

Step 2: Securing File Uploads

• Used secure_filename from werkzeug.utils to sanitize uploaded file names.
• Ensures only image files are accepted by validating file extensions.

Step 3: Containerizing the Application

• Dockerfile:

  FROM python:3.11-slim-buster
  WORKDIR /app
  COPY requirements.txt .
  RUN pip install --no-cache-dir -r requirements.txt
  COPY . .
  EXPOSE 5000
  CMD [ "python", "app.py" ]

• docker-compose.yml:

  version: '3.8'
  services:
    flask-app:
      build: .
      container_name: flask-app
      ports:
        - '5000:5000'
      environment:
        - FLASK_APP=app.py
        - FLASK_ENV=development
      volumes:
        - ./data:/app/data

Step 4: Adding a Regression Model Endpoint

• Endpoint: /regpredict
• Description:
  • Uses a pre-trained regression model (stored as a pickle file).
  • Predicts an employee's salary based on input features.

Step 5: Adding a Text Generation Endpoint

• Endpoint: /textgen
• Description:
  • Uses Hugging Face’s GPT pipeline to generate text based on a user-provided prompt.

Step 6: Adding a Home Route

• Endpoint: /home
• Description:
  • Serves as a documentation page explaining the project, its purpose, and deployment steps.

---

Deployment Instructions

1. Clone the repository:

   git clone https://github.com/mohamed-stifi/Flask-ML-Deployment.git
   cd Flask-ML-Deployment
   cd app

2. Build the Docker container:

   docker-compose up --build

3. Wait until the container completes setup.
4. Access the application at http://localhost:5000.
5. Use the following routes:
   • /predict: Object recognition.
   • /regpredict: Regression predictions.
   • /textgen: Text generation.
   • /home: Project overview and documentation.

---

Evaluation

• Functionality: All endpoints work as intended, demonstrating diverse AI capabilities.
• Security: File uploads sanitized using secure_filename.
• Portability: Application containerized for easy deployment across different environments.
• Scalability: Supports the addition of new models with minimal modifications.

---

Conclusion

This project successfully demonstrates a distributed automation pipeline for AI applications using Flask, Docker, and modern AI frameworks. The modular design allows easy expansion for future AI model deployments.