Drop-in encrypted Fairlearn metrics over CKKS. Same API surface; ciphertext arithmetic via TenSEAL or OpenFHE.

🔬 Drop in any ML model → get SHAP explainability, fairness audit & drift detection in seconds

AI-powered bias detection for datasets and ML models — with fairness metrics, natural language reports, and explainability tools.

Demo's of FairLearn and InterpretML as described in my article on responsible AI.

Reoffending-risk prediction with a Neo4j knowledge graph and a Fairlearn audit on the COMPAS dataset. A small auditable architecture demonstration.

End-to-end MLOps pipeline for heart disease classification, deployed on GKE with CI/CD, explainability, fairness testing, and drift detection.

Student Success Model (SSM)

Pipeline MLOps end-to-end para predição de churn em fintech, com governança (Fairlearn), tracking (MLflow), CI/CD (GitHub Actions), API (FastAPI), Docker e drift detection.

A comprehensive fairness analysis of the Boston Residents Jobs Policy compliance data using Fairlearn to detect and mitigate bias in construction project employment

End-to-end bias audit of healthcare ML models using MEPS dataset. Detects racial disparities, applies 4 mitigation techniques (AIF360), explains predictions (SHAP/LIME), and visualizes findings via Streamlit dashboard.

AI-powered LLM hallucination detection and responsible AI dashboard using Streamlit, FastAPI, Docker, SHAP, LIME, Fairlearn, and NLP pipelines.

AI Bias Detection & Mitigation Workbench: measure disparate impact with bootstrap confidence intervals, apply measured mitigations, and govern model sign-off. Fairness math validated against Fairlearn in CI.

Independent data privacy & AI governance assessment of a listed Indian NBFC's AI credit model — DPDP, DPIA, algorithmic bias audit, NIST AI RMF.

Explainable credit underwriting with SHAP local and global explanations, FairLearn fairness auditing, and FCRA-style adverse-action notice generation. XGBoost and LightGBM models served via a Streamlit dashboard.

Bias-aware machine learning system for fair credit risk assessment | 67% bias reduction | Fairlearn + SHAP

Reproducible CPU-only clinical risk prediction MLOps platform with calibration, conformal abstention, fairness, drift monitoring, and CI/CD.

Automated rubric-anchored response scorer with fairness audit and explainability — QWK 0.931, adverse-impact detection, attention-rollout highlights