Harley Nuel

Key Skills

Software

Appen

CVAT

HiveMind

Labelbox

Mercor

Mindrift

OneForma

Scale AI

Don't disclose

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Document

Image

Text

Top Label Types

Audio Recording

Classification

Object Detection

Prompt Response Writing SFT

RLHF

Segmentation

Transcription

Freelancer Overview

I am a data analyst and data annotation specialist with over 3 years of experience working on diverse projects involving data labeling, quality assurance, and AI training data preparation. My background includes hands-on annotation of text and image datasets for machine learning models, rigorous quality checks, and strict adherence to multi-step guidelines. I am highly skilled in Python, SQL, PowerBI, Excel, and JSON, and have worked extensively with large, complex datasets to ensure high data integrity and consistency. My experience spans domains such as business analytics, customer behavior, and public datasets, where I have built automated data pipelines, performed schema validation, and produced detailed technical documentation. I excel at translating complex data into actionable insights and ensuring that all training data meets the highest standards of accuracy and reliability. Detail-oriented AI Data Labeler and Annotation Specialist with 3 years of experience supporting machine learning pipelines across NLP, computer vision, and RLHF projects. Proven track record of delivering high-accuracy labeled datasets, maintaining annotation consistency across large-scale tasks, and collaborating with AI researchers to refine labeling guidelines. Experienced in text classification, entity recognition, image segmentation, bounding box annotation, and human feedback collection for large language model training.

Labeling Experience

Multilingual Toxicity Detection Labeler

LabelboxTextClassification

For the Multilingual Toxicity Detection Dataset, I labeled over 25,000 social media text samples in English and Pidgin for hate speech, harassment, and toxicity. I applied nuanced severity categorization, maintaining high individual auditing accuracy. My annotation work enabled a clean dataset release for model training. • Labeled data across multiple languages for broader coverage. • Achieved a 97.3% final accuracy score on audit evaluations. • Enhanced dataset quality with nuanced severity tiering. • Contributed to a high-fidelity, released training set.

For the Multilingual Toxicity Detection Dataset, I labeled over 25,000 social media text samples in English and Pidgin for hate speech, harassment, and toxicity. I applied nuanced severity categorization, maintaining high individual auditing accuracy. My annotation work enabled a clean dataset release for model training. • Labeled data across multiple languages for broader coverage. • Achieved a 97.3% final accuracy score on audit evaluations. • Enhanced dataset quality with nuanced severity tiering. • Contributed to a high-fidelity, released training set.

2022 - 2025

RLHF Response Ranking Project Annotator

Surge AITextRLHF

As part of the RLHF Response Ranking Project, I evaluated and ranked over 15,000 LLM-generated prompt-response pairs. I provided clear written justifications for ranking decisions aligned with help, harmlessness, and honesty benchmarks. These efforts supported model refinement and labeling guideline updates. • Identified model response failure patterns and reported them for QA review. • Wrote structured rationales supporting preference alignment. • Contributed to structured edge case documentation. • Enabled the adoption of improved guidelines by the annotation team.

As part of the RLHF Response Ranking Project, I evaluated and ranked over 15,000 LLM-generated prompt-response pairs. I provided clear written justifications for ranking decisions aligned with help, harmlessness, and honesty benchmarks. These efforts supported model refinement and labeling guideline updates. • Identified model response failure patterns and reported them for QA review. • Wrote structured rationales supporting preference alignment. • Contributed to structured edge case documentation. • Enabled the adoption of improved guidelines by the annotation team.

2022 - 2025

AI Data Annotation Specialist

LabelboxTextClassification

As an AI Data Annotation Specialist, I labeled more than 120,000 text samples supporting NLP projects such as sentiment analysis, NER, toxicity detection, and intent classification. I performed ranking and written preference justifications in RLHF workflows for large language model fine-tuning. My work contributed to maintaining a high inter-annotator agreement and exceptional QA scores. • Maintained an IAA score of 94% and quality above 96%. • Processed and validated JSON/JSONL datasets for ML pipeline readiness. • Recommended guideline clarifications that reduced disputes by 18%. • Met or exceeded daily throughput targets of 500–800 labeled items.

As an AI Data Annotation Specialist, I labeled more than 120,000 text samples supporting NLP projects such as sentiment analysis, NER, toxicity detection, and intent classification. I performed ranking and written preference justifications in RLHF workflows for large language model fine-tuning. My work contributed to maintaining a high inter-annotator agreement and exceptional QA scores. • Maintained an IAA score of 94% and quality above 96%. • Processed and validated JSON/JSONL datasets for ML pipeline readiness. • Recommended guideline clarifications that reduced disputes by 18%. • Met or exceeded daily throughput targets of 500–800 labeled items.

2022 - 2025

Web and Data Scraping

Don T DiscloseTextData Collection

Vibecode Specialist in Web Scraping & Data Extraction

Vibecode Specialist in Web Scraping & Data Extraction

2023 - 2024

Image & Video Annotation Analyst

CVATImageSegmentation

As an Image & Video Annotation Analyst, I annotated over 60,000 images and videos for autonomy and surveillance AI models. I performed bounding box and polygon segmentation for object detection and lane marking, maintaining high consistency with ground truth. My annotations supported model deployment in critical environments. • Achieved 97% alignment on golden set calibration for quality assurance. • Conducted temporal video annotation for activity recognition tasks. • Used CVAT to deliver pixel-accurate image segmentation. • Documented edge case decisions for model team traceability.

As an Image & Video Annotation Analyst, I annotated over 60,000 images and videos for autonomy and surveillance AI models. I performed bounding box and polygon segmentation for object detection and lane marking, maintaining high consistency with ground truth. My annotations supported model deployment in critical environments. • Achieved 97% alignment on golden set calibration for quality assurance. • Conducted temporal video annotation for activity recognition tasks. • Used CVAT to deliver pixel-accurate image segmentation. • Documented edge case decisions for model team traceability.

2021 - 2022

Show more

Education

Work History