DPA-MLOps

This is a tool to quickly spin up an MLFlow tracking server with a couple of clicks instead of worrying about the credentials, and backend addresses. Focus on training ML models while we take care of boring parts of MLOps.

How to use

Press Ctrl+Shift+P and search for 'DPA MLOps: Home'.

Press 'Settings' and update the fields with credentials given by IT team and save.

Now, either navigate the UI with 'DPA MLOps: Home' command or use 'MLOps: Quick Server' and 'MLOps: Stop Server' for a fast, no-non-sense experience

Features

1. One command to spin up a 'Quick Server', Save and modify your credentials in a GUI for ease of access.
2. Required conda environments will be created and requried packages will be automatically installed on the first run.
3. Compatible with both Windows and Linux

Requirements

Need to have anaconda or miniconda installed.

Extension Settings

You will be able to save and modify the login credentials that will be stored in your home directory. Port and IP of deployed tracking server can be modified.

Known Issues

Closing VS code will stop all ongoing servers. Has a complicated relation to ssh.

Release Notes

0.0.1

First release: Included faster no-UI experience in addition to UI

Working on features

1. Improving the UI
2. Bug fixes
3. Compatibility with mac os