home / research
Research
A core of boundary-focused systems security, with web and AI security alongside it.
Core research areas
reproducible · isolated · security-boundary systemsOS Security
Privilege separation, hardening and attack surface of general-purpose and minimal operating systems.
Kernel Security
Memory safety, syscall and driver interfaces, and exploit mitigations at the privilege boundary.
Isolation
Namespaces, sandboxes, hypervisors and enclaves: what they actually separate, and where they leak.
Air-Gapped Systems
Offline architectures, controlled data transfer and exposure to covert channels.
Reproducible Builds
Deterministic toolchains and verifiable artifacts, so what you run is what was reviewed.
Boot Security
Secure and measured boot, firmware trust and the chain from reset vector to userspace.
Trusted Computing
TPMs, attestation and roots of trust: proving system state instead of assuming it.
Beyond the core
web · AI · toolingWeb Security Research
Authentication and access control, SSRF, injection and framework-specific flaws in modern web stacks.
writing: /thoughtsAI Security Research
Prompt and tool-output injection, trust propagation across multi-agent systems, and unsafe model loading in AI infrastructure.
refs: OWASP LLM Top 10, MITRE ATLASWorkflow Tooling
Small tools I build for myself so research stays repeatable: triage, reproduction harnesses and report generation.