Standard-agnostic compliance checker engine. 8-layer NLP pipeline (TF-IDF, graph analysis, ensemble classification, fuzzy matching) that maps any safety standard to any work product and generates compliance reports. Electron desktop app included.

A portfolio & reference project demonstrating safety-critical development using DO-178C workflow. Includes planning docs, requirements, traceability, MISRA-C code, tests, and certification evidence.

This repository holds work completed by myself and the research group at McSCert to bring to life our idea of modelling safety engineering process to generate assurance cases directly from engineering process

Functional safety toolkit for AI-assisted development — ISO 26262, IEC 61508, IEC 60730, Automotive SPICE, MISRA-C:2012

Architecture-complete green-hydrogen validation repo: safety-gated bench testing, gas verification, carbon/source accounting, BOMs, assembly walkthroughs, Gankyil-A A/B tests, and evidence receipts.

This repository houses my PhD Thesis work

Concept-stage harbor-vessel retrofit architecture for berth-powered hydrogen generation from treated seawater, 350-bar storage, fuel-cell + battery hybrid power, layered safety/interlocks, and an auditable proof path for marine decarbonization decisions.

A multi-agent safety engineering framework that subjects systems to adversarial audit. Orchestrates specialized agents (Engineer, Psychologist, Physicist) to find process risks and human factors.

The PSA Control Standard defines governance controls for deployed agentic systems.

LUMINA-30: non-binding civilizational boundary framework for preserving effective human refusal authority before irreversible external impact from advanced AI systems.

Horizon-Eval: evaluation-integrity framework for portable long-horizon agent benchmarks, with QA gates, trajectory auditing, replayable run bundles, and safety-gap analysis.

A practical safety and recovery design package for AI / LLM-based control systems. Defines where AI can be used, where it must be stopped, and how systems recover.

Toy governance CLI demo: deny-by-default “danger actions” gated by signed, expiring capability leases (global revoke-all + nonce revoke) plus guarded memory quarantine. Simulation-only: no real network/shell/files. Includes tripwire + tests to prevent misuse.

A fast, elegant & free Event Tree Analysis (ETA) utility powered by 🤖 Godot and 🦀 Rust.

Agent-based simulation of emergency evacuation in university lecture halls — Master's Thesis, TU Clausthal 2025

Automatic classification of vehicle defect descriptions into component categories using TF-IDF and Logistic Regression, trained on 237,780 real NHTSA recall records. 98% accuracy on 9-class model.

Foundational architecture and doctrine for interoperable, autonomous cyber-physical production systems.

A fast, elegant & free ETA analysis utility

NLP pipeline for classifying unstructured "Near-Miss" (Hiyari-Hatto) reports. Uses TF-IDF & statistical classification to structure ambiguous safety data for risk analysis. ⚠️ 📊

Modular middleware for emotional safety, symbolic transparency, and real-time drift prevention. From bind-aware routing to EU-ready ethics, LLOYD Preserver protects behavior, motifs, and trust in language systems.