James Le

Key Skills

Software

Scale AI

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Image

Text

Video

Top Label Types

Classification

Data Collection

Diagnosis

Evaluation Rating

Function Calling

Mapping

Prompt Response Writing SFT

Question Answering

Relationship

Text Generation

Text Summarization

Freelancer Overview

I am an experienced AI training and data labeling specialist with a strong background in evaluating and annotating multimodal data, including text, audio, and technical domains. My work spans reviewing LLM outputs for accuracy, safety, and instruction-following, as well as designing and applying detailed rubrics for error classification, preference ranking, and quality assurance. With a PhD in Medical Physics, I bring scientific rigor and meticulous attention to detail to every project, ensuring high-quality, consistent labels and evaluations. I have hands-on experience with tools like Google Sheets, Excel, and transcription platforms, and have contributed to projects involving language data (English and Vietnamese), technical QA, and audio evaluation. I am fully equipped and available for remote, long-form AI tasks, and thrive in roles that demand precision, structured feedback, and collaborative problem-solving.

Labeling Experience

Labelbox (image labeling)

LabelboxImageClassification

Reviewed and debugged model responses (medical physics), providing structured error reports and quality feedback to improve training datasets.

Reviewed and debugged model responses (medical physics), providing structured error reports and quality feedback to improve training datasets.

2025

Outlier (labelling projects)

Scale AIDocumentQuestion Answering

Evaluated and debugged LLM outputs across physics, math, safety, and general reasoning tasks. Annotated, labeled, and validated training data to ensure accuracy, consistency, and guideline compliance. Designed and applied rubrics for scoring correctness, harmfulness, clarity, and instruction adherence. Performed preference ranking and pairwise comparisons to guide model optimization. Conducted audio QA: assessed background noise, clipping, speaker overlap, emotional tone, and labeling errors. Produced structured quality reports highlighting failure modes, edge cases, and mis-training risks. Reviewed and debugged model responses (physics, math, and general tasks), providing structured error reports and quality feedback to improve training datasets.

Evaluated and debugged LLM outputs across physics, math, safety, and general reasoning tasks. Annotated, labeled, and validated training data to ensure accuracy, consistency, and guideline compliance. Designed and applied rubrics for scoring correctness, harmfulness, clarity, and instruction adherence. Performed preference ranking and pairwise comparisons to guide model optimization. Conducted audio QA: assessed background noise, clipping, speaker overlap, emotional tone, and labeling errors. Produced structured quality reports highlighting failure modes, edge cases, and mis-training risks. Reviewed and debugged model responses (physics, math, and general tasks), providing structured error reports and quality feedback to improve training datasets.

2024

Education

Work History