Overview
What this challenge is about.
You receive a synthetic-but-realistic dataset of 25,000 past applicants with features (years of experience, education tier, prior role tags) and outcome labels (advanced past the screen / didn't). Train a logistic regression and a random forest as 'representative' production-like models. Then evaluate both for performance disparities across subgroup splits using simple group fairness metrics (false-positive-rate gap, true-positive-rate gap, demographic parity ratio). Run two simple mitigations (reweighting, threshold-per-group) and compare. Write a 3-page summary with a clear recommendation: ship, mitigate, or escalate to the client.
The Brief
What you'll do, and what you'll demonstrate.
Audit a hiring-screening classifier for subgroup performance gaps and recommend whether to ship, mitigate, or escalate.
Earning criteria — what you'll demonstrate
- Compute and interpret common group-fairness metrics on a real classifier
- Apply simple bias-mitigation techniques and measure their effect
- Recognize when a model's performance gap is large enough to block deployment
- Communicate audit findings to a non-technical client stakeholder
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
AI Safety Researcher
Running a fairness audit, comparing mitigations, and translating findings into a ship/mitigate/escalate call is exactly the work entry-level AI safety researchers do at consultancies and in-house responsible-AI teams.
This challenge sharpens
- fairness-metrics
- bias-mitigation
- model-evaluation
Data Scientist
Pairing model training with subgroup analysis and clear stakeholder communication is the modern data scientist's job description in any regulated industry.
This challenge sharpens
- logistic-regression
- random-forest
- model-evaluation
Applied AI Scientist
Quantifying the accuracy/fairness trade-off honestly and recommending a path forward mirrors the daily work of applied AI scientists supporting product teams.
This challenge sharpens
- bias-mitigation
- fairness-metrics
- python