akihiro-inui/ml_service_template

Template for Machine Learning Services using FastAPI, PyTorch and MLFlow

Machine Learning Service Template

Template for the projects with FastAPI to service predictions with Machine Learning models.

This repo contains

1. Fast API based API server
2. ML Training Code infrastructure with MLFlow

Features

Built-in Swagger Document for API

If you managed to run the application, you should be able to access to localhost:80 and see the Swagger documentation like this.

Model Training with MLFlow

This repo contains the infrastructure to train the model with MLFlow. You can use the MLFlow UI to track the training process and the model performance.

Deployment

Serving Pattern

It follows Web Single Pattern architecture for the API server due to its advantage of the maintainability and compatibility with the API server and the model.