Thanks for helping make automated hiring fairer and more transparent.
- Contributions are under Apache 2.0 (
LICENSE/NOTICE). - Never commit real candidate data. Use
app/services/seed_data.py(synthetic, deterministic). Demographic fields are audit-only signals. - Keep the method defensible: if you change scoring or attribution, update
docs/01-methodology.mdand add/adjust tests.
pip install -r requirements.txt
pytest -q
python -m app.services.pipeline_demo- A kernel/tree SHAP explainer for a non-linear reference model.
- Impact-ratio (four-fifths rule) reporting on top of the bias score.
- An adapter that ingests a real ATS's decision payload.
- Rendering the Mermaid diagrams to committed SVG/PNG.
- Add/adjust a test for any behavioural change (
pytest -qmust stay green). - Note which endpoint or method your change touches.