deepsense.ai Project Template

Useful template to bootstrap new professional data science and python projects.

Documentation

Table of Contents

• What is it?
• What are the benefits?
• Getting started

What is it?

Its intended use is to generate basic, most common configuration - however each team and developer is encouraged to modify it for its special needs.

It is a result of our experiences with building data science projects and is a part of our internal best practices, however it is not a silver bullet and should be treated as a starting point for your project. Especially some settings might be less/more restrictive than you needs but we believe it is better to start with a good baseline and modify it later than to start from scratch.

What are the benefits?

Generated project consists of:

1. Basic python package structure:
   • pyproject.toml - central configuration for build system, package metadata, dependencies, and tool configurations (including linters and formatters)
   • a very minimal python code + example test
2. pre-commit hooks:
   • ruff - fast linter and formatter that replaces the functionality of:
     • black - code formatting
     • flake8 - style and linting
     • pycln - unused imports cleanup
     • isort - import sorting
     • pylint - static code analysis
     • pyupgrade - code modernization for given python version
     • bandit - security issue detection - required for SOC 2 Certification
   • mypy - static type checking
   • jupytext - (optional) syncs jupyter notebooks to plain python files
3. Sphinx documentation:
   • basic preconfigured documentation template
   • recommended extensions
   • page with list of autogenerated third party python packages list with licenses
4. Basic script to create venv
5. Minimal README.md file
6. Preconfigured semantic versioning with bump2version
7. Dockerfile for pre-commit image
8. Gitlab integration (default, optional):
   • linter stage (pre-commit run --all)
   • tests (pytest) + code coverage
   • license checks of installed packages
   • building and hosting documentation on GitLab Pages
   • building package and uploading to private GitLab Package registry
   • security: trivy - required for SOC 2 Certification
   • steps to rebuild linter docker image
9. GitHub integration (optional):
   • linter stage (pre-commit run --all)
   • tests (pytest) + code coverage
   • license checks of installed packages
   • building and hosting documentation on GitHub Pages (if enabled)
   • testing if package can be build
   • security: trivy - required for SOC 2 Certification, but is reporting only
10. Other less important files (more configurations, .gitignore etc.)

Most up-to date descriptions, tips and explanations are in the documentation.

Getting started

Generate project template locally:

Install cookiecutter (at least >=2.1.1 version) first and then point it to this repository.

    $ pip install cookiecutter>=2.1.1

Cookiecutter will ask you set of questions so it can generate customized project.

    $ cookiecutter ds-template/
    client_name [ds]: Client Name
    project_name [default]: Sunglass
    repo_name [client-name-sunglass]:
    ...

How to initialize new repository with the template:

Firstly, you need to create a new project. The name should be of the following convention:

<client_name>-<project_name>

Execute the following steps then:

Approach 1 (clone empty):

# clone empty repository to repo_name
$ git clone <GIT-SSH>
# install cookiecutter if not yet installed
$ pip install cookiecutter
# generate cookiecutter with --force and ensure the repo_name is set to the same name as directory you cloned git repository to.
$ cookiecutter -f [email protected]:deepsense-ai/ds-template.git
# finally, add all files, commit and push.
$ git add .
$ git commit -m "Initialize repository with default project template"
$ git push origin

Approach 2 (initialize git locally and push to remote):

# install cookiecutter if not yet installed
$ pip install cookiecutter
# generate project
$ cookiecutter [email protected]:deepsense-ai/ds-template.git
# enter created directory
$ cd <project-name>
# now we need to connect it to repository (assuming empty repository)
$ git init
$ git remote add origin <GIT-SSH>
$ git fetch
$ git checkout -t origin/main
# finally, add all files, commit and push.
$ git add .
$ git commit -m "Initialize repository with default project template"
$ git push --set-upstream origin main
$ git push origin