Operational AI governance
How inventories, roles, risk classification, evidence, oversight cadence and decision records become an operating model rather than a compliance document.
L2ET Research is organized around long-horizon tracks where AI, cyber, assurance, resilience and trust converge. Research interests span AI assurance, agentic systems, cybersecurity governance, post-quantum transition, scientific computing, quantum information, and selected cyber-physical/autonomous systems questions.
How inventories, roles, risk classification, evidence, oversight cadence and decision records become an operating model rather than a compliance document.
How systems with retrieval, memory, tools and delegated authority can be constrained, observed, escalated and evaluated before operational dependence forms.
How prompt channels, RAG exposure, tool permissioning, identity boundaries and incident evidence change the security perimeter.
Cryptographic discovery, long-lived confidentiality, protocol pressure, supplier evidence and migration sequencing before transition urgency removes room to manoeuvre.
Research at the boundary of formal reasoning, computational structure and domain-specific AI, including physics-aware and mathematically constrained systems.
How organizations can produce inspectable evidence of control, resilience and accountability as AI moves into high-consequence workflows.