LogS-model-API

A FastAPI endpoint for a machine learning model to predict LogS.

Local development

The following commands will setup an environment where you can run and test the application locally:

git clone [email protected]:jonswain/ml-model-api.git
cd logs-model-api
mamba env create -f envs/dev.yml
conda activate LogS-model-API
code .

Running a local uvicorn ASGI web server

A local server, running on the default port (8000), can be started as follows:

uvicorn app.main:app

Once started, you can browse to http://localhost:8000 to test the app.

Running a local gunicorn WSGI web server in docker

A docker container running gunicorn can be started locally using docker compose:

docker compose up -d

The base image handles sensible gunicorn configuration options.

The backend service can be accessed at port 5000 (http://localhost:5000)

Directory structure

├── README.md               <- The top-level README for developers using this project.
├── app
│   ├── __init__.py         <- Makes LogS-model-API a Python module.
│   ├── config.py           <- Config data for machine learning model API.
│   ├── main.py             <- API functions for hosting.
│   ├── post_processing.py  <- Scripts for processing predictions.
│   ├── prediction.py       <- Scripts for making predictions.
│   └── pydantic_classes.py <- Pydantic classes for API data validation.
│
├── envs                    <- Conda environments.
│   └── dev.yml             <- Conda environment for development.
│
├── model                   <- Directory to place your machine learning model in.
│
├── tests                   <- Test functions for deployment.
│           
├── .dockerignore
│           
├── .gitignore
│           
├── Dockerfile              <- Docker compose for development.
│
└── requirements.txt        <- Requirements file for dependencies in Docker container.