Independent researcher focused on semantic consistency boundaries and proxy-alignment drift in frontier LLMs. Currently investigating the intersection of multi-step reasoning depth and internal logic-anchoring. Technical inquiries: kurmaobito@gmail.com