Abbie Trewin-bugeja

Key Skills

Software

CVAT

Data Annotation Tech

Labelbox

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Image

Text

Top Label Types

Segmentation

RLHF

Evaluation Rating

Audio Recording

Text Generation

Prompt Response Writing SFT

Bounding Box

Object Detection

Classification

Transcription

Emotion Recognition

Red Teaming

Freelancer Overview

I am an experienced AI Trainer and Data Annotator with a strong background in image, audio, and text data labeling across domains such as medical, legal, and aviation. My work includes detailed image annotation using tools like LabelBox and CVAT, transcribing and validating audio with precise formatting, generating and evaluating prompts for model robustness, and performing pairwise comparisons to assess model outputs for truthfulness and instruction following. I am adept at identifying and documenting model vulnerabilities, designing adversarial prompts, and ensuring high-quality, consistent datasets. My technical skills span Excel (complex formulas, PivotTables), SAP, and various proprietary labeling platforms, and I bring a keen eye for detail, process improvement, and cross-functional collaboration to every project.

Labeling Experience

AI Trainer / Data Annotator at Alignerr

LabelboxAudioClassificationEmotion Recognition

I evaluated AI voice models to select genre-appropriate voices and documented preferences and rationale. I transcribed ATC audio with precise formatting using the NATO alphabet and airline/waypoint cues. My work ensured the highest standard of dataset quality for audio-based AI model training. • Conducted genre-specific voice evaluation and documentation • Performed exacting ATC audio transcriptions for dataset integrity • Employed audio data annotation methodologies • Supported voice model optimization for AI performance

I evaluated AI voice models to select genre-appropriate voices and documented preferences and rationale. I transcribed ATC audio with precise formatting using the NATO alphabet and airline/waypoint cues. My work ensured the highest standard of dataset quality for audio-based AI model training. • Conducted genre-specific voice evaluation and documentation • Performed exacting ATC audio transcriptions for dataset integrity • Employed audio data annotation methodologies • Supported voice model optimization for AI performance

2024

AI Trainer / Data Annotator at Outlier

LabelboxImageBounding BoxClassification

I performed detailed image annotation, focusing on labeling people, clothing, objects, landmarks, and styles to improve model training outcomes. I created reverse image prompts and generated factual and deceptive statements for model evaluation tasks. I validated and corrected AI audio transcripts, and conducted comparisons of multimodal prompt-responses to assess truthfulness and formatting. • Enhanced dataset quality through meticulous labeling and analysis • Developed strategies for evaluating AI understanding and robustness • Increased transcription accuracy using timestamped audio validation • Employed medical, English language, and management domain expertise

I performed detailed image annotation, focusing on labeling people, clothing, objects, landmarks, and styles to improve model training outcomes. I created reverse image prompts and generated factual and deceptive statements for model evaluation tasks. I validated and corrected AI audio transcripts, and conducted comparisons of multimodal prompt-responses to assess truthfulness and formatting. • Enhanced dataset quality through meticulous labeling and analysis • Developed strategies for evaluating AI understanding and robustness • Increased transcription accuracy using timestamped audio validation • Employed medical, English language, and management domain expertise

2024

AI Trainer / Data Annotator at Data Annotation Tech

Data Annotation TechTextClassificationRLHF

I reviewed AI-generated responses, verified factual claims, and evaluated accuracy and safety based on detailed rubrics. I assessed stylistic consistency, tone alignment, and tested AI spreadsheet functions for correctness. I conducted adversarial prompt creation, flagged vulnerabilities, and recommended improvements for model safety and reliability. • Developed and executed prompt tests for AI understanding • Annotated and categorized images • Documented reproducible safety and bias issues • Enhanced formula accuracy in data-handling tasks • Writing prompt and response pairs/multi-turn conversations demonstrating ideal responses

I reviewed AI-generated responses, verified factual claims, and evaluated accuracy and safety based on detailed rubrics. I assessed stylistic consistency, tone alignment, and tested AI spreadsheet functions for correctness. I conducted adversarial prompt creation, flagged vulnerabilities, and recommended improvements for model safety and reliability. • Developed and executed prompt tests for AI understanding • Annotated and categorized images • Documented reproducible safety and bias issues • Enhanced formula accuracy in data-handling tasks • Writing prompt and response pairs/multi-turn conversations demonstrating ideal responses

2022 - 2024

Education

Work History