Qundong C.

Key Skills

Software

No software listed

Top Subject Matter

大语言模型rlhf/指令优化与对话数据质检 Domain Expertise

指令对齐nlp标注、模型评测与偏好数据构建 Domain Expertise

多模态数据标注与nlp（情感分类/ner） Domain Expertise

Top Data Types

Text

Image

Audio

Top Task Types

RLHF

Prompt + Response Writing (SFT)

Classification

Freelancer Overview

高级人工智能训练师（GLM系列RLHF数据标注与质检）. Brings 7+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal and Proprietary Tooling. Education includes Bachelor of Science, Beijing University of Posts and Telecommunications (2020) and Level 4 Professional Certificate, Ministry of Human Resources and Social Security (2022). AI-training focus includes data types such as Text and Image and labeling workflows including RLHF, Prompt + Response Writing (SFT), and Classification.

Labeling Experience

高级人工智能训练师（GLM系列RLHF数据标注与质检）

TextRLHF

Responsible for GLM series LLM RLHF human feedback data collection and quality control, producing over 50,000 high-quality annotated entries. Designed and optimized prompt engineering strategies to improve accuracy on Chinese reasoning tasks. Led a small annotation team to construct multi-turn dialogue datasets with strong consistency and shorter delivery cycles. • GLM RLHF data collection and quality assurance • Prompt engineering for Chinese reasoning accuracy improvement • Multi-round dialogue dataset construction with consistency targets • Participation in internal annotation standards and QC process formulation

2023 - Present

人工智能训练师（NLP方向：指令对齐与评测）

TextPrompt + Response Writing (SFT)

Conducted instruction alignment annotation for the early version of Wenxin Yiyan, covering multiple task types such as summarization, question answering, and code generation. Performed manual evaluation of model outputs with rating dimensions including helpfulness, harmlessness, and honesty. Contributed to building preference datasets for multi-turn dialogue to support DPO training and analyzed hallucination patterns for categorization. • Instruction alignment annotation across 10+ task types • Manual quality evaluation (helpfulness/harmlessness/honesty) • Preference dataset construction for DPO training • Hallucination case reproduction, analysis, and pattern extraction

2021 - 2023

数据标注员（多模态数据与NLP标注）

ImageClassification

Completed annotation and review work for multi-modal datasets including images, text, and audio at high daily throughput. Participated in NLP labeling projects such as sentiment classification and named entity recognition (NER). Ensured dataset delivery volumes met project targets through continuous quality checks. • Multi-modal data annotation and审核（图像/文本/语音） • Text sentiment classification labeling • NER labeling for named entity extraction • Review and quality control to meet high throughput targets

2020 - 2021

Education

Work History