Paul E.

Key Skills

Software

CVAT

Other

Top Subject Matter

LLM training data (RLHF preference ranking) and AI alignment

Legal document NLP annotation (NER, coreference) and intent classification

Autonomous vehicle perception dataset labeling

Top Data Types

Text

Document

Image

Audio

Video

Top Task Types

RLHF

Entity (NER) Classification

Bounding Box

Transcription

Diagnosis

Classification

Tracking

Freelancer Overview

Senior Data Labeling Specialist — DataForce by TransPerfect (AI Data Labeling). Brings 1+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Internal, Proprietary Tooling, and CVAT. Education includes Bachelor's Degree, St. Paul's University (2025). AI-training focus includes data types such as Text, Document, and Image and labeling workflows including RLHF, Entity (NER) Classification, and Bounding Box.

Labeling Experience

Senior Data Labeling Specialist — DataForce by TransPerfect

DocumentEntity (NER) Classification

Carried out entity extraction and coreference resolution on legal documents with performance measured against a gold standard. Improved annotator alignment by designing annotation guidelines for a complex multi-label intent task, reducing disagreement from 18% to 5% over 3 sprints. Built Python scripts to automate annotation quality checks and reduce manual QA effort. • Performed entity extraction and coreference resolution on legal texts • Reduced inter-annotator disagreement via guideline authoring and iterations • Automated QA verification with Python scripts for faster review • Used agreement and metric tracking to monitor labeling quality

2022 - Present

Senior Data Labeling Specialist — DataForce by TransPerfect (AI Data Labeling)

TextRLHF

Produced RLHF training data by annotating 400K+ preference pairs and ranking model outputs using rubrics. Ensured quality and consistency across 6 languages with a 97.3% project QA pass rate and high inter-annotator agreement. Applied evaluation criteria spanning helpfulness, harmlessness, and honesty for alignment and fine-tuning workflows. • Annotated prompt-response pairs for constitutional AI fine-tuning datasets • Ranked outputs using multi-dimensional safety and appropriateness rubrics • Maintained QA pass-rate targets through systematic checks • Calibrated annotators via weekly calibration sessions and dashboards

2022 - Present

AI Annotation Analyst — Appen Limited

OtherAudioTranscription

Labeled 120K+ audio clips for speech recognition fine-tuning with transcription plus accent and noise-level tagging. Ensured label consistency to support model training and downstream evaluation. Participated in additional NLP annotation tasks including search relevance and project adjudication. • Completed transcription and accent tagging for speech datasets • Labeled noise-level attributes for training robustness • Maintained a 99.1% accuracy rate during a 12-month audit window • Coordinated multi-annotator adjudication for toxicity classification outputs

2020 - 2021

AI Annotation Analyst — Appen Limited

CVATImageBounding Box

Labeled over 800K image bounding boxes and segmentation masks for autonomous vehicle perception datasets. Used CVAT for structured computer-vision annotation and maintained high personal accuracy through ongoing quality audits. Contributed to perception dataset readiness for model training and evaluation. • Annotated bounding boxes and segmentation masks in CVAT • Supported dataset creation for autonomous driving perception • Maintained accuracy targets during quality audits • Followed labeling protocols for consistent dataset generation

2020 - 2021

Data Collection & Labeling Associate — Samasource (Sama)

OtherVideoTracking

Participated in video object tracking annotation projects for retail analytics AI, labeling 200+ hours of footage. Applied tracking labeling practices to generate temporal datasets for computer-vision model training. Followed project protocols to ensure continuity and consistency across frames. • Produced video object tracking labels for retail analytics AI • Labeled 200+ hours of video footage • Maintained temporal consistency for tracking annotations • Supported dataset creation for downstream CV model training

2019 - 2020

Show more

Education

Work History