A beginner-friendly AI Governance & Risk Toolkit — risk register, governance templates, and audit-ready workflows for early-stage AI teams.

Data Trust Engineering (DTE) is a vendor-neutral, engineering-first approach to building trusted, Data, Analytics and AI-ready data systems. This repo hosts the Manifesto, Patterns, and the Trust Dashboard MVP.

Four Tests Standard (4TS) - Vendor-neutral specification for verifiable AI governance

Formales Protokoll zur epistemischen Output-Governance von KI-Systemen.

Supporting materials for “Building Governable ML Models with R,” presented at posit::conf 2025

Regime-based evaluation framework for financial NLP stability. Implements chronological cross-validation, semantic drift quantification via Jensen-Shannon divergence, and multi-faceted robustness profiling. Replicates Sun et al.'s (2025) methodology with modular, auditable Python codebase.

Deterministic stress testing of levered UK rental cashflows under rate and vacancy shocks.

Ethical AI governance framework for multi-model alignment, integrity, and enterprise oversight.

This project demonstrates how autonomous AI agents can collaborate under strict validation, guardrails, and human-in-the-loop approval to analyze financial data and produce executive-ready reports, designed for regulated environments such as banking and financial services.

Production-ready MSME Credit Risk Pipeline (V3.0). Solved critical data integrity issues (target/scaling) for 47% AUC lift (0.88). Model implements a hard-cutoff policy based on DPD/Utilization, ensuring portfolio PD drops below the 3.75% break-even threshold.

Drift observability architecture for Databricks Delta Lake — detects data & model drifts, builds PSI visualizations, and exports governance telemetry for Responsible AI.

Professional AI Security Assurance portfolio demonstrating model supply-chain security, LLM red teaming, static analysis, SBOM validation, risk classification, and governance-aligned AI safety workflows.

A platform that makes your domain model executable and shared across humans, systems and AI agents, so nothing is guessed and work stops being re-done. One explicit, documented model becomes the single ground truth that cuts governance overhead, removes ambiguity, and lets AI act with accuracy instead of approximation.