AI system discovery and risk mapping for agentic architectures. Generates CycloneDX AIBOMs with topology and risk classification.

AI system registry and governance tracker with EU AI Act risk classification, impact assessments, and a compliance dashboard — built for legal and compliance teams.

Clean-room project exploring reviewable, testable, accountable human-AI collaboration workflows.

A machine learning tool that predicts the risk of loan default using borrower financial data and outputs model probabilities and classifications from logistic regression, XGBoost, and a neural network.

Governance control plane for EU AI Act compliance — system registry, risk classification, evidence RAG, eval gates, data-contract drift monitoring, and audit-ready evidence packs

Claude Code plugin for structured change management — risk classification (P0-P3), multi-domain collaborative analysis, adversarial spec review, implementation verification, and business impact analysis.

Injury prediction model using machine learning to analyze factors like workload, player metrics, and environmental conditions. It identifies injury risk patterns early, enabling preventive actions, improved training decisions, and reduced injury occurrence in athletes.

SQL-Analyse eines Kreditportfolios (32.581 Datensätze) mit Risikoklassifikation in MySQL, Projektarbeit IHK Data Analyst (DataSmart Point, 2026)

Guardrail API for detecting personalized mental-health advice with explainable safe/unsafe decisions. Docker-first, FastAPI, Cloud Run-ready, with auth, rate limits, and cost controls.

Computer-vision crowd monitoring pipeline built on FIDTM localization, with polygon zone analysis, rule-based risk classification, refined analytics, a Streamlit dashboard, and an LLM-based decision-support assistant. Bachelor thesis, German University in Cairo

A FastAPI service that classifies AI use cases against EU AI Act risk tiers: Unacceptable, High, Limited, or Minimal risk.

LLM complaint insights: Structured summaries, risk tags, and entity extraction on CFPB consumer complaints — built with OpenAI + LangChain, MLflow, Streamlit.

NSCLC Radiomics: Survival Time prediction using CT-extracted features and clinical data.

Deterministic EU AI Act classifier with source-backed review rows, obligation links, dossier artifacts, and blocked external actions.

Enterprise IAM governance pipeline — maps 75 applications to IAM products, classifies risk, detects governance gaps, and produces management dashboards with remediation queues and department scorecards.

Experimental Claude Code skill for solo builders: structured prompts and reference checklists that surface the invisible production layer (security, GDPR, scaling, bus factor) of any build. No warranty; not legal or security advice.

120-Tool Autonomous AI Command Center - EU AI Act Compliance - Multi-Agent MCP + A2A Protocol - Nexus Control Center

Reference repository outlining protocol-level risk and liability classification concepts; taxonomy reference, not insurance or financial advice.

This project uses a public vehicle trajectory dataset (proxy) to simulate real-time driving signals.

POC de monitoramento agrícola com dados espaciais, ESP32/DHT22, Machine Learning e Streamlit para classificação de risco e recomendações agrícolas.