DVC Project Design Patterns (WIP)

Introduction

Welcome to the "DVC Project Design Patterns" community project! This initiative is dedicated to exploring, documenting, and sharing best practices and design patterns for using Data Version Control (DVC), an open-source tool for managing and versioning data and machine learning models.

What You Will Find Here

Design Pattern Catalogue: A collection of design patterns for common scenarios encountered in machine learning projects, such as handling large datasets, managing experiments, and parallel processing.
Best Practices: Tips and tricks for optimizing DVC workflows, managing dependencies, and ensuring reproducibility.
Real-World Examples: Practical implementations of these patterns in real-world projects, demonstrating their application and benefits.

Who Should Participate

Data Scientists & ML Engineers: Improve your workflow with efficient DVC practices.
Team Leads & Managers: Understand how to structure ML projects for better manageability and collaboration.
Open Source Enthusiasts: Contribute by sharing your own experiences, patterns, and best practices.

Installation

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
export PYTHONPATH=$PWD

Patterns

1 - Parallel Stages

Pattern	Description	Example
Parallel Stages	Run multiple independed stages of a pipeline concurrently	pipelines/parallel-stages

graph TD;
    data --> train_rf["Random Forest"]:::focusStyle;
    data --> train_lr["Linear Regression"]:::focusStyle;
    train_rf --> evaluate;
    train_lr --> evaluate;

    classDef focusStyle fill:#f49f,stroke-width:1px

2 - Parallel Pipelines

Pattern	Description	Example
Parallel Pipelines	Run multiple pipelines concurrently	pipelines/parallel-pipelines

graph TD;
    subgraph model_rf ["train_rf/dvc.yaml"]
        train_rf --> test_rf;
    end 
    subgraph model_lr ["train_lr/dvc.yaml"]
        train_lr --> test_lr;
    end 

    1(data/dvc.yaml) --> model_rf(model_rf/dvc.yaml):::focusStyle;
    1(data/dvc.yaml) --> model_lr(model_lr):::focusStyle;

    model_rf --> 4(evaluate/dvc.yaml);
    model_lr --> 4(evaluate/dvc.yaml);

    classDef focusStyle fill:#f49f,stroke-width:1px,rx:5,ry:5

Contributing

This is a community-driven project, and your contributions are vital to its success. You can contribute in several ways:

Share a Pattern: Document a design pattern you've found useful in your projects.
Contribute with Examples: Implement an existing pattern in a new context and share your findings.
Improve Documentation: Help refine the explanations, add clarity, or fix errors.
Spread the Word: Share this project with your network to grow the community.

mnrozhkov / dvc-patterns