丁立恒

Key Skills

Software

No software listed

Top Subject Matter

AI Evaluation

Multi-Agent Workflows

Product Development

Top Data Types

Text

Document

Image

Top Task Types

RLHF

Evaluation/Rating

Computer Programming/Coding

Prompt + Response Writing (SFT)

Text Summarization

Question Answering

Red Teaming

Fine-tuning

Freelancer Overview

AI Trainer and Mandarin AI Evaluator with over 2 years of hands-on experience in AI response evaluation, prompt engineering, and quality assurance. Specialized in comparing model outputs, detecting hallucinations, assessing instruction-following, and providing structured RLHF-style feedback. Strong expertise in reviewing AI-generated code, UI/UX implementations, product documents, and Chinese-language content. Experienced in AI-assisted workflows using tools like ChatGPT, Cursor, DeepSeek, and OpenCode across real software projects. Native Mandarin speaker with professional English proficiency, capable of delivering high-quality bilingual evaluation and feedback.

Labeling Experience

Founder / Product Builder — AI-Assisted Product Intelligence

Text

Built and evaluated AI-generated product intelligence and opportunity reports using public user signals and demand data. Provided structured assessments of pain point summaries, supporting evidence, and actionable insights generated by language models and workflow agents. Evaluated clarity, factual accuracy, and value of English/Chinese content in iterative AI/LLM QA cycles. • Oversaw AI-driven insight extraction and report refinement workflows. • Evaluated and rated generated documentation and reports for quality assurance. • Managed bilingual AI report review and feedback processes. • Utilized proprietary and external toolchains such as ChatGPT, DeepSeek, and Codex.

2025 - Present

AI Trainer & Evaluator (Independent Projects)

TextEvaluation/Rating

Reviewed and evaluated AI-generated responses for accuracy, usefulness, completeness, tone, and instruction-following. Compared outputs from multiple large language models, identified hallucinations and failure modes. Provided structured RLHF-style feedback in both English and Chinese. Also conducted code review, UI/UX quality assessment, and bilingual content evaluation for AI-assisted software projects.

2024 - Present

Education

Work History