
FairHire AI is a comprehensive solution developed by team CodeXAI to address the critical issue of bias in AI-driven recruitment decisions. Unlike existing tools that typically flag bias only after a model is built, this platform proactively prevents bias by suggesting "Neutral Language" rewrites for job descriptions using Google Gemini. The platform offers several key features and capabilities:Data Auditing: It scans CSV and SQL datasets for underrepresentation of specific demographic groups and measures "Disparate Impact" to ensure minority groups are not unfairly filtered out. Mitigation Tools: It includes a Gemini Rewrite Engine to re-phrase biased job requirements (e.g., changing "coding ninja" to "expert software engineer") and provides scripts to "re-weight" training data to fix identified issues. Visualization: A Fairness Dashboard provides real-time "Bias Score" visualizations and heatmaps to help HR teams see where models might be making discriminatory mistakes. Architecture: The solution is built with a Python/Flask backend, a JavaScript frontend, and utilizes MongoDB for maintaining a transparent audit trail of hiring trends. Future Development: Plans for the project include real-time monitoring for model drift, integration with global legal frameworks like the EU AI Act, and expanding bias detection to include image and voice recognition.
10 May 2026