Independent builder exploring how persistent state, layered rules, and long-horizon memory change LLM behavior over time. Author of Self-becoming, an open-source 5-month experiment in functional self-continuity — not a lab researcher, just someone trying to make the question testable.