Supervision and Mentorship

Agentic AI for Adaptive Cyber Physical System Defence

Supervising doctoral research on a novel agentic AI framework for autonomous incident response in cyber-physical systems, applying agentic AI and neuro-symbolic techniques to correlate IT and OT telemetry and reconstruct multi-stage attacks against the MITRE ATT&CK for ICS framework. The framework generates dynamic, standards-aligned response playbooks (NIST SP 800-61, STIX/TAXII), reaches multi-agent consensus with human-in-the-loop approval before acting in physical environments, and is evaluated on a Cyber-Digital-Twin testbed against realistic attack simulations.

Agentic Cognitive Firewall Open Source Project

Contributed to ACF-SDK, a framework-agnostic security layer that defends LLM agents against direct and indirect prompt injection by intercepting user input, RAG context, tool calls, and memory reads and writes. Built an attack benchmark to evaluate and expand the SDK's attack-detection coverage.

Agentic Cognitive Firewall Open Source Project

Contributed to ACF-SDK, a framework-agnostic security layer that defends LLM agents against direct and indirect prompt injection by intercepting user input, RAG context, tool calls, and memory reads and writes. Focused on improving prompt pattern detection through the SDK's sentence-transformer-based semantic scanner.

Bridging Intent and Actuation in LLM-Based Robotic Systems

Hosted a Marie Curie PhD researcher (University College Cork) on secondment at Nimbus Research Centre, supervising development of a framework to capture the gap between intent and actuation in LLM-based robots.

Mobile Threat Intelligence Hub: On-Device Smishing Detection via Edge SLMs

Supervising an undergraduate Software Engineering final year project at the Informatics Institute of Technology (IIT), Sri Lanka, developing on-device smishing detection using edge-deployed small language models.