Research
Pathfinder AI — a single-agent system implementing the four-layer agentic-generative VAPT model from my systematic literature review, evaluated against a SWaT-inspired Modbus testbed.
→ research log// PhD candidate · security analyst · GRC
Building agentic AI
for the systems
that actually matter.
I research autonomous vulnerability assessment in operational technology environments — the unglamorous infrastructure that runs water treatment, power grids, and factory floors. Day to day I work as a security analyst in an MSP environment, with a focus on cloud security and GRC.
Experience
Security analyst work across managed client estates: Microsoft 365 and Azure security, incident triage, governance, and compliance. Azure-native, ISO 27001-fluent.
→ track recordProjects
Independent work — a multi-PLC OT testbed, an OpenAI Model Craft Challenge submission (51% loss reduction over baseline), and writing on agentic systems.
→ portfolio // flagship
Pathfinder AI
A four-layer agentic VAPT prototype: observe / reason / decide / act / store. Local-first (Ollama + Qwen), evaluated on an OT testbed.
// paperAgentic and Generative AI for Autonomous VAPT
Co-authored systematic analysis. Targeted at Computers & Security.
// researchClawbot — agent insider threat
How a locally-deployed AI agent becomes an insider threat vector through prompt manipulation alone.