Setup and version your ML project

with MLV-tools and DVC

Stéphanie Bracaloni & Sarah Diot-Girard

About Us

Sarah Diot-Girard
Data Scientist since 2012
Interested in DataOps and Ethics

@SgdJlbl

Stéphanie Bracaloni
Software Engineer since 2013
Automation and Code Quality

@sbracaloni

Major issues:

Versioning (Data & Jupyter Notebooks)
Automate the pipeline execution
Reproduce & Share experiments

Setup a project using MLV-tools

Easily create DVC steps from command line pipeline steps

gen_dvc -i ./script.py -o ./commands/script_dvc

Setup a project using MLV-tools

Convert Jupyter Notebooks to
executable and configurable Python3 scripts

ipynb_to_python -n ./notebook.ipynb -o ./script.py

Setup a project using MLV-tools

Convert Jupyter Notebooks to DVC steps in one command

ipynb_to_dvc -n ./notebook.ipynb -o ./commands/script_dvc


[Tutorial time]

Clone MLV-tools tutorial repository

Resources:
mlv-tools-tutorial/resources/setup_project

How to:
mlv-tools-tutorial/tutorial/setup_project.md

We are waiting for your Pull Requests!

github.com/peopledoc/ml-versioning-tools
github.com/peopledoc/mlv-tools-tutorial

Fonts: White Rabbit by Matthew Welch
Capsuula by Henrich Fichna
Icons made by Smashicons, Good Ware, Designmodo, Freepik from www.flaticon.com; license CC 3.0 BY

Contact Us !

sarah_diot-girard@ultimatesoftware.com

stephanie_bracaloni@ultimatesoftware.com

Time Optimization

Tutorial Part1: build docker images

cd mlv-tools-tutorial/resources/dvc_playground
docker-compose build

Tutorial Part2: build docker image (Optional)

(Can also be run on your computer)

                ./resources/setup_project/docker/run.sh