Pipeline that ingests raw credit data, profiles quality with WhyLogs, applies cleaning and encoding, and stores versioned feature sets. Addresses EU AI Act Articles 10 and 12.
Pipeline that trains LightGBM models with bias-aware techniques, evaluates performance across demographic groups, and conducts comprehensive risk assessment. Addresses Articles 9, 11, and 15.
Pipeline that implements human oversight gates, deploys models to Modal, generates SBOM and Annex IV documentation, and sets up post-market monitoring. Addresses Articles 14, 17, and 18.
Credit Scorer is a production-ready credit scoring system that demonstrates full EU AI Act compliance for high-risk AI systems. Built with comprehensive MLOps pipelines, it automatically generates all required technical documentation, ensures robust risk management, and maintains complete audit trails with versioned artifacts.
Credit Scorer transforms loan application data into compliant credit scoring predictions while automatically generating all EU AI Act compliance artifacts. It processes sensitive financial data through automated pipelines that ensure data quality, fairness across demographic groups, and comprehensive documentation for regulatory inspection.
