Peter O'connor

Key Skills

Software

Labelbox

Mercor

Telus

Top Subject Matter

Large Language Models

NLP Tasks

Geospatial Search and Mapping Evaluation

Top Data Types

Document

Image

Text

Audio

Top Task Types

Data Collection

Emotion Recognition

Prompt + Response Writing (SFT)

Question Answering

Text Summarization

Audio Recording

Freelancer Overview

LLM Training, Prompt Design & Evaluation Specialist. Brings 13+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Mercor, Telus, and Internal. AI-training focus includes data types such as Text, Geospatial, and Tiled Imagery and labeling workflows including Prompt + Response Writing (SFT), Evaluation, and Rating.

Labeling Experience

Search & Mapping Data Analyst

Telus

Rated and annotated location search results for relevance, accuracy, and user intent using proprietary mapping tools. Applied structured guidelines to evaluate business listings, routes, and geographic data across large datasets. Produced high-quality labeled datasets to enhance AI-driven mapping and search algorithms. • Utilized Telus proprietary mapping platform for data labeling. • Compared multiple candidate results for intent alignment. • Maintained attention to consistency and annotation detail. • Supported the development of improved map-based AI systems.

Rated and annotated location search results for relevance, accuracy, and user intent using proprietary mapping tools. Applied structured guidelines to evaluate business listings, routes, and geographic data across large datasets. Produced high-quality labeled datasets to enhance AI-driven mapping and search algorithms. • Utilized Telus proprietary mapping platform for data labeling. • Compared multiple candidate results for intent alignment. • Maintained attention to consistency and annotation detail. • Supported the development of improved map-based AI systems.

2026 - Present

LLM Training, Prompt Design & Evaluation Specialist

MercorTextPrompt Response Writing SFT

Developed prompts and evaluation rubrics to improve LLM reasoning and behavior in a reinforcement learning environment. Ranked and assessed multiple generated model outputs using structured scoring methods. Authored clear written justifications for all ranking decisions, focusing on accuracy and instruction adherence. • Created evaluation scaffolds for supervised fine-tuning (SFT). • Assessed outputs for summarization, question answering, and synthesis tasks. • Used internal/proprietary Mercor platform for labeling. • Ensured consistent application of quality standards across varied domains.

Developed prompts and evaluation rubrics to improve LLM reasoning and behavior in a reinforcement learning environment. Ranked and assessed multiple generated model outputs using structured scoring methods. Authored clear written justifications for all ranking decisions, focusing on accuracy and instruction adherence. • Created evaluation scaffolds for supervised fine-tuning (SFT). • Assessed outputs for summarization, question answering, and synthesis tasks. • Used internal/proprietary Mercor platform for labeling. • Ensured consistent application of quality standards across varied domains.

2026 - Present

TTS Voice Data Contributor

AudioAudio Recording

Recorded high-quality dialogue audio for internal TTS (Text-to-Speech) model training, following strict technical guidelines. Delivered multiple character voices and emotional directions based on provided scripts and briefs. Contributed expressive, noise-minimized voice samples for machine learning data pipelines. • Produced over 3 hours of audio for training datasets. • Followed 24-bit/48 kHz WAV standards, ensuring consistency. • Interpreted scripts for narration and emotive clarity. • Supported improvements in TTS accuracy and naturalness.

Recorded high-quality dialogue audio for internal TTS (Text-to-Speech) model training, following strict technical guidelines. Delivered multiple character voices and emotional directions based on provided scripts and briefs. Contributed expressive, noise-minimized voice samples for machine learning data pipelines. • Produced over 3 hours of audio for training datasets. • Followed 24-bit/48 kHz WAV standards, ensuring consistency. • Interpreted scripts for narration and emotive clarity. • Supported improvements in TTS accuracy and naturalness.

2026 - 2026

Education

Work History