decision-policy

A decision-safety lab for loan approval: trains a baseline classifier, calibrates probabilities (ECE/Brier), sweeps confidence thresholds to build a coverage, quality frontier and outputs a defensible abstention policy (auto-decide vs review). Includes a Streamlit dashboard for report cards, triage UI, and data quality checks.

POC de CRM Analytics focada em modelagem de propensão à resposta, com análise exploratória orientada à decisão, engenharia de atributos e preparação de dados para profit targeting.

Calibrated GBT (isotonic XGBoost) scores first-scan promise-miss risk on public Olist data. Exception handling policy; LangGraph/MCP agent executes it in simulation. Judged on business impact (actions, OTIF, spend), not intermediate model performance. Production: Cloud Run, Feast, MLflow, canary, monitoring, human-gated promote.

POC antifraude baseada em Machine Learning, que demonstra como estruturar modelos como motores de decisão, com avaliação temporal honesta, baseline sólido e comparação entre algoritmos.

Enable reliable loan decisions by calibrating model scores into defendable actions: auto-approve, auto-reject, or manual review policies.