Uu W.

Key Skills

Software

Label Studio

Top Subject Matter

ML Domain Expertise

Nlp Domain Expertise

GIS Domain Expertise

Top Data Types

Text

Image

Geospatial Tiled Imagery

Top Task Types

Prompt + Response Writing (SFT)

Classification

Text Generation

RLHF

Fine-tuning

Freelancer Overview

LLM Data Annotator & QA Dataset Creator (MedRAFT Chinese Medical QA). Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal and Proprietary Tooling. Education includes Master of Science, National University of Singapore (2025) and Bachelor of Science, Zhejiang University (2025). AI-training focus includes data types such as Text and labeling workflows including Prompt + Response Writing (SFT).

Labeling Experience

LLM Data Pipeline Builder & Annotator (MiniMind Reproduction)

TextPrompt + Response Writing (SFT)

I built an end-to-end LLM pipeline in PyTorch focused on data cleaning, supervised fine-tuning, and reinforcement learning from human feedback. This involved structuring raw textual data, formatting prompts and responses, and crafting supervised learning datasets for AI model training. The work also included workflow design for various training stages and adaptation methods. • Processed and cleaned raw textual data for model training. • Generated prompt-response pairs for supervised fine-tuning. • Supported RLHF/DPO stages with prompt curation and quality checking. • Facilitated effective training dataset creation for Transformer-based models.

2026 - 2026

LLM Data Annotator & QA Dataset Creator (MedRAFT Chinese Medical QA)

TextPrompt + Response Writing (SFT)

I contributed to the creation of Chinese medical QA training data for an LLM fine-tuning project. I constructed and supervised question-answer pairs, including teacher-supervised and distractor-augmented samples, to enhance the model's language understanding and robustness. My work enabled efficient data normalization, prompt formatting, and evaluation of model outputs for academic research. • Designed and generated 1,199 teacher-supervised QA samples. • Created 1,195 distractor-augmented instances to improve model robustness. • Implemented evaluation pipelines for retrieval and hallucination analysis. • Assisted in dataset preparation and quality assurance for LLM training.

2025 - 2025

Education

Work History