Joy Buehler

Key Skills

Software

Internal/Proprietary Tooling

Telus

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Text

Video

Top Label Types

Action Recognition

Audio Recording

Classification

Emotion Recognition

Evaluation Rating

Fine Tuning

Object Detection

Prompt Response Writing SFT

Question Answering

Red Teaming

Relationship

RLHF

Segmentation

Text Generation

Text Summarization

Transcription

Freelancer Overview

I bring hands-on experience as an AI data trainer, where I evaluated outputs for large language models using RLHF and SFT methods to improve generative AI performance, including a pilot contract with NVIDIA and additional NDA-restricted projects across multiple industries. My background in tech-driven operations, team training, and process optimization allows me to deliver high-quality data annotation and feedback within fast-paced, detail-oriented environments. I am skilled in a range of workflow and collaboration tools such as Asana, Slack, Notion, and Google Workplace, and have worked remotely with cross-functional teams. My diverse experience—from hospitality operations to health tech and AI—enables me to quickly grasp domain-specific requirements and ensure data integrity for machine learning applications.

Labeling Experience

AI Trainer

Internal Proprietary ToolingTextQuestion AnsweringText Generation

Worked as an AI trainer and annotator on a multilingual language-AI project administered by Translated for an enterprise client (NVIDIA). Scope of work included evaluating and annotating AI-generated text used to train and refine large language models, with emphasis on linguistic accuracy, semantic correctness, and strict adherence to project guidelines. Tasks involved reviewing model outputs, applying structured labels, correcting or refining responses, and performing quality-focused text annotation such as entity tagging, meaning alignment, and error identification. All work was completed within a controlled annotation platform using client-defined rubrics. Quality was measured through accuracy thresholds, consistency checks, and ongoing performance monitoring, with continued access to projects dependent on meeting required standards.

Worked as an AI trainer and annotator on a multilingual language-AI project administered by Translated for an enterprise client (NVIDIA). Scope of work included evaluating and annotating AI-generated text used to train and refine large language models, with emphasis on linguistic accuracy, semantic correctness, and strict adherence to project guidelines. Tasks involved reviewing model outputs, applying structured labels, correcting or refining responses, and performing quality-focused text annotation such as entity tagging, meaning alignment, and error identification. All work was completed within a controlled annotation platform using client-defined rubrics. Quality was measured through accuracy thresholds, consistency checks, and ongoing performance monitoring, with continued access to projects dependent on meeting required standards.

2024

AI Data Labeling

TelusVideoObject DetectionAction Recognition

Performed audio annotation work for RWS on short-form video clips, focused on identifying and labeling discrete sound events. Tasks included tagging sounds with precise start and end timestamps, assigning standardized labels, and providing brief descriptions in accordance with detailed project guidelines. Work was completed within a controlled annotation platform with strict requirements for accuracy, consistency, and edge-case handling. Quality was measured through ongoing QA review, guideline adherence, and correction of flagged items, with expectations to adapt quickly to updated instructions. The role required sustained attention to detail, careful auditory judgment, and consistent application of labeling standards across varied audio environments.

Performed audio annotation work for RWS on short-form video clips, focused on identifying and labeling discrete sound events. Tasks included tagging sounds with precise start and end timestamps, assigning standardized labels, and providing brief descriptions in accordance with detailed project guidelines. Work was completed within a controlled annotation platform with strict requirements for accuracy, consistency, and edge-case handling. Quality was measured through ongoing QA review, guideline adherence, and correction of flagged items, with expectations to adapt quickly to updated instructions. The role required sustained attention to detail, careful auditory judgment, and consistent application of labeling standards across varied audio environments.

2025 - 2025

Data Annotator

Internal Proprietary ToolingTextClassificationQuestion Answering

The primary responsibility was to review, label, and evaluate AI outputs against project-specific criteria. Tasks ranged from comparing multiple model responses and fact-checking or rating them for accuracy, relevance, tone, and adherence to instructions, to tagging data points and applying detailed annotations based on provided guidelines, with the goal of improving large language models through human judgment and correction. Clear written instructions and task guidelines were a core part of the process, and workers were expected to follow these closely to ensure high consistency and quality in labeling, since pay and continued access to projects depended on meeting accuracy standards.

The primary responsibility was to review, label, and evaluate AI outputs against project-specific criteria. Tasks ranged from comparing multiple model responses and fact-checking or rating them for accuracy, relevance, tone, and adherence to instructions, to tagging data points and applying detailed annotations based on provided guidelines, with the goal of improving large language models through human judgment and correction. Clear written instructions and task guidelines were a core part of the process, and workers were expected to follow these closely to ensure high consistency and quality in labeling, since pay and continued access to projects depended on meeting accuracy standards.

2023 - 2024

Education

Work History