016-data-scientist/workspace/t1/github/surge/CONTRIBUTING.md

Contributing

Thanks for your interest. Surge is a small open-source project; PRs and issues welcome.

Development setup

git clone https://github.com/<your-fork>/surge.git
cd surge
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev,api]"

You'll need a few free API keys in .env (see .env.example):

• EIA_API_KEY — register at https://www.eia.gov/opendata/register.php
• NREL_API_KEY (optional) — https://developer.nrel.gov/signup/
• NCDC_TOKEN (optional) — https://www.ncdc.noaa.gov/cdo-web/token

Running tests

pytest

Running the API locally

# pull enough data to run inference
python -m surge.ingest --days 90

# download a model checkpoint (separate — see docs/models.md)
# ...

uvicorn surge.api.main:app --reload

Style

• ruff check src tests must pass
• mypy src must pass
• Match surrounding style — small surface area, keep dependencies minimal.

What goes in git vs. what doesn't

• Source code: git
• Tests and fixtures: git
• Benchmarks / results: experiments/results.tsv is gitignored — publish to HF datasets
• Model weights: gitignored — publish to HF Hub
• Parquet data: gitignored — can be rebuilt via python -m surge.ingest

Publishing

• Dataset snapshots → huggingface.co/datasets/surge-grid/*
• Model checkpoints → huggingface.co/surge-grid/*