fairlearn

Why we included this project

Fairness is one of those things teams know they should check but rarely have a concrete way to check. Fairlearn gives you the tools: metrics that compare a model's behavior across demographic groups, and mitigation algorithms that adjust training or post-processing to reduce the gaps it finds. It's a Python package that fits into the workflow you already have, so fairness checks run alongside your normal evaluation rather than as a separate exercise. For teams accountable for how their models treat people, that turns a policy concern into something measurable.