Yusheng S.

Key Skills

Software

Label Studio

Top Subject Matter

LLM evaluation

prompt-response assessment

Chinese/English content

Top Data Types

Text

Image

Top Task Types

Text Generation

Question Answering

Text Summarization

Prompt + Response Writing (SFT)

Freelancer Overview

AI Product Researcher / LLM Evaluation Specialist. Core strengths include Internal and Proprietary Tooling. AI-training focus includes data types such as Text and labeling workflows including Evaluation and Rating.

Labeling Experience

Product Researcher — Personal Knowledge & Evidence-Based AI Systems

Text

I led product research for evidence-based AI systems, designing knowledge retrieval workflows focused on citation and reliable recall. I defined product requirements and annotation schemes to ensure grounded responses, source traceability, and hallucination reduction. My work included designing evaluation protocols and structured labeling for knowledge recall and trust. • Created principles for managing not-found cases and reducing hallucinations • Evaluated hybrid search and citation quality in retrieval-augmented generation • Designed annotation protocols for evidence and reasoning • Analyzed user needs for improved knowledge recall and accessibility

2024 - Present

AI Research & Evaluation Contributor

Text

At Confidential AI Research Projects, I contributed to AI-directed data evaluation and annotation initiatives. My key responsibilities included output review, prompt testing, response comparison, and model behavior analysis. I conducted evaluation and refinement of AI responses for instruction adherence and user utility. • Analyzed hallucination risk and weak source grounding in model outputs • Evaluated ambiguous reasoning and long-context reliability • Assisted with prompt effectiveness and user intent satisfaction • Supported data annotation workflows for internal research

2024 - Present

AI Product Researcher / LLM Evaluation Specialist

Text

As an AI Product Researcher and LLM Evaluation Specialist, I evaluated AI-generated prompts and responses in both Chinese and English. My tasks included rating output for factual accuracy, clarity, usefulness, and instruction-following, as well as reviewing prompt design and identifying hallucinations and alignment failures. I developed practical evaluation criteria for AI systems covering evidence traceability, source grounding, and reliability. • Compared multiple AI outputs to identify vague reasoning and unsupported claims • Reviewed naturalness, cultural appropriateness, and coherence of responses • Developed structured annotation schemes as per evaluation guidelines • Focused on user intent alignment and safety policy compliance

2024 - Present

Education

Work History