Elim Q.

Key Skills

Software

Other

Top Subject Matter

Chinese Classical Dance Tutoring/Education

LLM Decision-Making/Audit

Top Data Types

Video

Text

Image

Top Task Types

Text Generation

Evaluation/Rating

Data Collection

Computer Programming/Coding

Object Detection

Question Answering

Text Summarization

Classification

Freelancer Overview

AI-Powered Chinese Classical Dance Tutoring Agent - LLM Workflow and Evaluation. Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Master of Science, University of Oxford (2025) and Bachelor of Arts, Shanghai University (2025). AI-training focus includes data types such as Video and Text and labeling workflows including Text Generation, Evaluation, and Rating.

Labeling Experience

Interpretable Steering for LLM Decision-Making using Activation Engineering

OtherText

Built an interpretable framework to analyze and steer LLM decision-making using activation engineering and representation analysis. Conducted large-scale evaluation by integrating over 73,000 World Values Survey samples and running 2,000 persona-based prompts to test demographic and alignment biases. Applied semantic vector steering and regression analysis to evaluate and neutralize LLM biases in environment versus economy decision tasks. • Utilized LLMs for social simulation and bias analysis using text data inputs. • Designed and executed quantitative evaluation prompts and ratings. • Used PCA and vector steering to assess model decisions. • Generated insights on alignment between AI output and human patterns.

2026 - 2026

AI-Powered Chinese Classical Dance Tutoring Agent - LLM Workflow and Evaluation

OtherVideoText Generation

Developed an AI-powered tutoring agent capable of breaking down Chinese classical dance videos into step-by-step movement explanations. Created a structured LLM workflow to generate detailed instructional feedback, technical highlights, and common mistakes for beginner and non-professional learners. Applied prompt engineering and evaluation processes to improve instructional clarity, output consistency, and movement interpretation quality. • Designed video-to-text workflows using LLMs for educational purposes. • Focused on extracting actionable dance steps and feedback from video data. • Evaluated and improved generated explanations to ensure clarity. • Utilized modular engineering and LLM integration throughout the process.

2026 - 2026

Education

Work History