During this career exploration phase, I dedicated 200+ hours to hands-on evaluation of frontier LLMs (ChatGPT, Claude, DeepSeek, Character.AI, Woebot, Wysa), building conversational AI prototypes and systematically assessing AI outputs for role-play consistency and emotional authenticity.
I also conducted in-depth research on digital mental health and cross-border brand intelligence, positioning myself at the intersection of content analysis and AI product innovation. This period strengthened my ability to identify subtle failures in AI-generated content — a skill directly transferable to AI training data evaluation and quality assurance.