NHU Y NGUYEN

I’m a results-driven AI Trainer (English - Vietnamese) focused on improving large language model outputs for factual accuracy, coherence, relevance, and safety. In my part-time freelance role, I evaluate AI-generated responses and help teams reduce issues that harm user trust.

I detect hallucinations, bias, and logical inconsistencies in model outputs, then develop and refine prompts to better align responses with user intent. I also create reproducible test cases and document error patterns so improvements are measurable and repeatable.

Alongside training work, I’ve built an AI Response Evaluation & Prompt Testing personal project, where I systematically evaluated 100+ AI-generated responses across multiple domains. I designed structured evaluation criteria (accuracy, fluency, relevance, safety), tested edge cases, and proposed changes to improve clarity and factual correctness.

I also strengthen my work with hands-on data practices: I’ve labeled datasets for text classification, sentiment analysis, and intent-focused tasks, ensuring annotation consistency and quality. With a strong background in linguistics and English teaching, I bring a careful, data-informed mindset to prompt engineering and response evaluation.