George Livingstone

Key Skills

Software

Internal/Proprietary Tooling

Top Subject Matter

Special Education

Audio Production and Recording

Top Data Types

Text

Document

Audio

Top Task Types

Text Summarization

Question Answering

Object Detection

Classification

Bounding Box

Prompt Response Writing SFT

Freelancer Overview

During my time at Outlier, I specialized in refining Large Language Models through Reinforcement Learning from Human Feedback (RLHF) and complex data annotation. My primary focus involved performing high-level response comparisons and side-by-side evaluations to rank model outputs based on technical accuracy, instruction following, and adherence to safety guidelines. I was responsible for identifying subtle "hallucinations" and logical inconsistencies, ensuring the models provided reliable, factually sound information. Beyond simple labeling, I crafted diverse, high-complexity prompts designed to stress-test model capabilities in areas like creative synthesis and multi-step reasoning. What sets my experience apart is the marriage of analytical rigor with a high degree of linguistic nuance. I developed a sharp eye for comparative ranking, often working within strict rubrics to prioritize helpfulness and tone while maintaining technical precision. My background allowed me to approach data labeling not just as a repetitive task, but as an iterative training process, consistently delivering high-quality feedback that directly influenced the model's behavioral alignment. This role required a unique blend of rapid research skills, critical thinking, and the ability to pivot between varied subject matters while maintaining a high throughput of verified data.

Labeling Experience

Education

Work History