m∞

Moo provides a stupidly easy way to deploy or invoke your ML models in cloud native environment like kubernetes. The auth and runtime components of Moo are greatly inspired by micro, the next generation microservice framework.

Moo is under active development, get ready to say hello to it 🙂

Architecture

5f86f43174dbf624_edit_262853693974995.jpg

Quick view

mooctl is the command line tool used to interact with Moo. Here are possible commands for a quick start according to the current design.

Deploy a model

mooctl run https://github.com/mivinci/mnist --replicas=3

Load a model

mooctl load mivinci/mnist --gpu_id=1 --forever

Call a model

mooctl call mivinci/mnist --input [email protected]/path/to/image

Documentation

All documentation will be open-sourced after the first stable release of Moo.

About the name

M stands for machine-learning models and ∞ for no reason but to be 🆒.

GitHub

View Github