About

Systems Mindset // Security Research

I am a security researcher and founding engineer focused on AI/ML systems, offensive security, and trust-boundary analysis.

My work centers on understanding how modern autonomous systems fail across model behavior, agent orchestration, parser logic, and distributed infrastructure. I approach AI systems not as abstract black boxes, but as complex stateful environments with real security assumptions, attack surfaces, and architectural weaknesses.

I spend most of my time building research labs, developing deterministic PoCs, analyzing alignment behavior, and studying how subtle systems assumptions evolve into real-world security impact.

Whether researching alignment abliteration techniques, auditing multi-agent workflows, or analyzing semantic trust-boundary drift in distributed applications, I focus on technically rigorous and transparent security research.

Research Toolbox

ModelsQwen, Llama, Mistral

LibrariesPyTorch, Transformers, LangChain, LangGraph, Hugging Face

SecurityBurp Suite, Whitebox Pentesting, Threat Modeling, OWASP

InfrastructurePython, Docker, FastAPI, Linux, Git

Core Research Interests

AI/LLM Security

Mechanistic Interpretability

Autonomous Agent Systems

Trust-Boundary Failures

Parser & Application Security

Offensive Security Research

Research Areas & Ongoing Work

1. AI/ML Security Research

Researching alignment behavior, refusal localization, mechanistic interpretability, and autonomous AI-agent security. Built and published Qwen3.5 alignment abliteration research artifacts using activation tracing and low-rank weight-space analysis.

2. Offensive Security & Whitebox Analysis

Conducting whitebox security research across distributed systems, application security, parser logic, and trust-boundary enforcement. Building deterministic PoCs and security labs focused on semantic drift, orchestration failures, and authorization inconsistencies.

3. Autonomous Security Systems

Building autonomous offensive-security infrastructure and multi-agent workflows at SecNode. Working on orchestration pipelines involving LangGraph, Model Context Protocol (MCP) integrations, and adaptive multi-model routing.

4. Coordinated Vulnerability Disclosure

Participating in coordinated vulnerability disclosure involving AI-agent systems, parser vulnerabilities, multi-tenant architectures, and application security. Submitted reports through programs including NASA's Vulnerability Disclosure Program via Bugcrowd.