Projects-Portfolio

Project standards

Every project in this portfolio follows the same rules, so anyone — a recruiter, a hiring manager, or future-me — can drop into any folder and know exactly where things are.

The rules

1. One project = one folder, under data/, ai-ml/, it-ops/, or experiments/, with a kebab-case name (data/energy-demand-forecast).
2. Every new project follows the same layout — the structure below (package code in src/, tests in tests/, raw data gitignored) is what the finished projects in this repo demonstrate. Copy an existing project folder as a starting point and rename it.
3. Python projects use pyproject.toml, a src/ package layout, and pytest.
4. No secrets. Real credentials live in .env (gitignored); the committed .env.example shows only the shape.
5. No raw data in git. data/raw/ is gitignored. Small, licence-clean samples may be committed under data/samples/ with provenance notes in the project README.
6. Every project has a README: problem → approach → results → how to run it.
7. CI must pass. pytest runs for every project on every push (see .github/workflows/ci.yml).

What a finished project looks like

data/my-project/
├── README.md            # problem, approach, results, quick start
├── pyproject.toml       # metadata + deps + pytest config
├── .env.example         # environment shape only, never values
├── src/my_project/      # package code
├── tests/               # pytest suite
├── notebooks/           # exploration / analysis (data & ai-ml projects)
└── data/
    ├── raw/             # gitignored — generated or downloaded on setup
    └── samples/         # optional: small committed samples for demos

This site is open source. Improve this page.