I conduct independent AI safety research focused on identifying protocol-level interaction risks in deployed systems and designing bounded, consent-preserving mitigations. My work emphasizes reproducibility, scope discipline, and practical implementation constraints rather than speculative claims or advocacy.