Hui Z.

Key Skills

Software

Don't disclose

Top Subject Matter

AI/LLM safety evaluation and alignment across scientific and behavioral domains

Legal Services & Contract Review

Regulatory Compliance & Risk Analysis

Top Data Types

Text

Document

Top Task Types

Bounding Box

Red Teaming

RLHF

Evaluation/Rating

Text Generation

Entity (NER) Classification

Freelancer Overview

Cross-domain AI evaluation expert with proven capability in model alignment, prompt engineering review, adversarial testing / red-teaming, and RLHF quality assurance. Combines deep technical domain knowledge (civil engineering, quantum mechanics, cognitive science) with rigorous evaluation methodology to assess AI outputs across reasoning, factual accuracy, and safety dimensions. Especially skilled at identifying edge cases and subtle failure modes that single-domain evaluators miss.

Labeling Experience

Cross-domain AI evaluation expert (alignment, prompt engineering review, adversarial testing, red-teaming validation)

Don't discloseText

Led cross-domain evaluation activities emphasizing AI alignment, model safety, and adversarial robustness testing through systematic assessment of model outputs. • Defined evaluation criteria spanning technical accuracy, logical coherence, and human-level cognitive alignment • Performed red-teaming-style validation to surface failure modes such as jailbreak susceptibility • Supported prompt engineering review to improve reliability and reduce unsafe behaviors • Assessed responses for alignment with scientific/engineering/philosophical and behavioral expectations.

2009 - 2024

Education

Work History