AI safety researcher. Interested in adversarial robustness, LLM evaluation, and cryptographic audit trails.