ML web service on FastAPI

This repository contains the files to build your own Machine Learning web application!

In this example we use NLP sentiment-analysis model from Hugging Face Hub. However the described code structure can be used for any machine learning problem.

Local development

# Create a virtual environment
python3.11 -m venv env

# Activate the virtual environment
source env/bin/activate

# Install/upgrade dependencies
pip install -U -e .
pip install -U -e .[dev]

# (Optional) Code formatting
make pretty

# Run tests for ml code
make test_ml

# Run app
uvicorn app.app:app --host 0.0.0.0 --port 8080

# Deactivate the virtual environment
deactivate

Run app in docker container

docker build -t ml-app .
docker run -p 80:80 ml-app

Run tests for the app

Run the following commands while docker container is running (in other terminal).

source env/bin/activate
make test_app

deactivate