Getting Started

Prerequisites

Install Docker

Follow the Docker installation guide for your system's requirements.

Install awslocal

This package provides the awslocal command, which is a thin wrapper around the AWS command line interface for use with LocalStack.

pip install "awscli-local[ver2]"

Install LocalStack

# MacOS
brew install localstack/tap/localstack-cli

# PyPI (MacOS, Windows, Linux)
python3 -m pip install localstack

Running the Application

Option 1: Using Docker Compose

# Build and run the containers
docker-compose build --no-cache
docker-compose up -d

# Note: You might need to manually restart Celery due to a known bug

Option 2: Running Services Individually

# Start Redis
docker run --name redis_container -p 6379:6379 redis/redis-stack-server

# Start PostgreSQL
docker run -d --name postgres-container \
    -e POSTGRES_DB=document_classifier \
    -e POSTGRES_USER=postgres \
    -e POSTGRES_PASSWORD=postgres \
    -p 5432:5432 postgres

# Additional Services:
# - LocalStack creates a docker image after installation
# - Flask can be started via VSCode debugger
# - Celery can be started with: celery -A src.tasks worker --pool=solo --loglevel=info

Usage Notes

Development Recommendations

• It's recommended to run services individually instead of using docker-compose for testing:
  • Better visibility of changes and logs
  • Avoid potential LocalStack docker issues
  • The project has been tested and works fine locally

Running the Application

1. Start the services using either method above
2. Access the application:
   • Docker Compose: localhost:8000
   • VSCode debugger: Custom port (e.g., Flask default 5000)
3. Send requests to /classify_file with file payload
4. Monitor progress:
   • Watch Celery terminal output
   • Poll /task_status/{task_id} endpoint

Architecture Notes

The project uses Celery+Redis for the following reasons:

• Immediate request acceptance and storage in Redis broker
• Asynchronous processing by workers
• Non-blocking /classify_file endpoint
• Results can be checked regularly for long-running tasks

Potential Improvements

• Reduce task duration by:
  • Allocating more resources (CPU/GPU)
  • Using models trained for specific document types
  • These options can be discussed based on company needs

---

Note: This solution demonstrates asynchronous processing capabilities while maintaining system responsiveness. Thank you for the opportunity!