I build AI-native systems for resilience, security, and decision-making
My work sits at the intersection of cyber architecture, simulation, incident response, and agentic AI. I use small, public builds as laboratories for bigger questions: how do we test AI systems, score decisions, simulate organisations, and make resilience measurable?
Current focus
- Deterministic simulation for AI-native decision systems
- Incident response readiness and cyber resilience
- Multi-agent coordination under real-world constraints
- Privacy and identity-exposure analysis with synthetic ground truth
Builds
Society-scale disaster response simulation. A multi-agent society of Qwen AI agents coordinates under pressure — triaging casualties, routing resources, and maintaining coherence when infrastructure fails.
bluntmachetti.github.io/aftershock ↗Incident-response training gym with deterministic scoring and auditable replay. Practice cyber incidents under realistic conditions. Every run is scored, logged, and replayable.
bluntmachetti.github.io/incidentgym ↗Agentic organisation and economy simulation. Multi-agent economies where agents negotiate, trade, form alliances, and fail — a testbed for governance mechanisms.
blog.redoubtlabs.dev/arena ↗Deterministic synthetic identity graphs for testing privacy, PII-extraction, and entity-resolution systems. Every record is unmistakably fake; every benchmark frozen and checksummed. Open source on PyPI.
github.com/bluntmachetti/synthworld ↗Product and research lab for AI-native resilience systems. Enterprise digital twins, operational resilience modelling, and decision simulation.
redoubtlabs.io ↗Enterprise Architecture Decision Simulator. Operational resilience modelling and enterprise-scale decision simulation. Selected private work.
Coming soonIdentity-exposure platform: maps what the public internet knows about a person and how it connects. SynthWorld is its open-source ground-truth harness.
Coming soon