mlreef brand

Install self-managed MLREEF

aka Nautilus

Installing MLReef on-premises

you can download, install and maintain your own MLReef instance

Want us to host your instance, no package required? Try MLReef SaaS today

Installing guide for Nautilus

This is the recommended method for getting started. The Linux packages are mature, scalable, and are used today on

Linux installation is quick to install, easy to upgrade and contains features to enhance realiability. Install via a single package (docker image) that bundles all the different services and tools required to run MLReef. At least 4GB of RAM is recommended

Installing MLReef On Premises

The best way to run MLReef on your own on-premises infrastructure is the MLReef Nautilus package. Nautilus is a single docker image containing everything necessary to create machine learning projects and run ML workloads.

Nautilus Contains:

  • MLReef Management Serive
  • Postgres
  • Gitlab for hosting Git repositories
  • Gitlab Runners for running Machine Learning workloads
  • API Gateway


In order to run MLReef Nautilus locally you just have to execute the followingdocker run command. This will start up a local instance of mlreef with persistent docker vilumes namedmlreef-opt, mlreef-etc, andmlreefdb-opt containing all user data.

docker run -it --rm --detach --name mlreef \
--volume /var/run/docker.sock:/var/run/docker.sock \
--volume mlreef-opt:/var/opt/gitlab   \
--volume mlreef-etc:/etc/gitlab       \
--volume mlreefdb-opt:/var/opt/mlreef \
--publish 2022:22                     \
--publish 80:80                       \
--publish 8081:8081                   \
--publish 5050:5050                   \
--publish 10080:10080                 \
--publish 6000:6000                   \

The container comes up with a default runner running on same docker network on localhost