Stephanie D.

Key Skills

Software

Appen

Telus

Top Subject Matter

Conversational AI and LLM response evaluation for safety

Tone Domain Expertise

and factuality

Top Data Types

Text

Image

Top Task Types

RLHF

Segmentation

Point/Key Point

Freelancer Overview

Senior AI Trainer / Data Specialist (LLM RLHF training & evaluation). Core strengths include Appen and Telus. Education includes Master of Engineering, Laurentian University (2024) and Bachelor of Science, University of Calabar (2022). AI-training focus includes data types such as Text and Image and labeling workflows including RLHF, Segmentation, and Point.

Labeling Experience

Senior AI Trainer / Data Specialist (computer vision segmentation labeling)

AppenImageSegmentation

Performed computer-vision semantic segmentation labeling for autonomous-vehicle related training datasets. Produced structured segmentation outputs as part of multi-class vision data preparation workflows. Worked with taxonomy and labeling guidelines to maintain consistency for downstream model training. • Completed semantic segmentation labeling for autonomous vehicle training • Produced segmentation annotations for computer vision datasets • Collaborated on taxonomy and labeling guideline refinement • Supported consistent labels for evolving project needs

2021 - 2023

Senior AI Trainer / Data Specialist (LLM RLHF training & evaluation)

AppenTextRLHF

Provided RLHF-based evaluation and ranking for conversational LLM outputs focused on safety, tone, and factual correctness. Optimized conversational AI agent behavior by using human feedback signals in the RLHF workflow. Contributed to model requirement evolution by applying labeling guidelines during iterative training cycles. • Ranked LLM responses for safety, tone, and factual correctness • Performed RLHF to optimize conversational AI agents • Applied labeling/taxonomy refinements with engineers • Ensured alignment to evolving model requirements through guideline updates

2021 - 2023

Data Annotator & Content Analyst (multimodal/dense labeling)

TelusImagePoint/Key Point

Labeled and analyzed multimodal datasets to improve search relevance and image recognition. Applied detailed gesture-related annotations including bounding boxes, polylines, and keypoint labels for training data. Maintained high annotation quality through auditing and peer review processes. • Applied precision bounding boxes, polylines, and keypoint annotations for gesture recognition • Labeled dense datasets for search relevance and image recognition • Analyzed multimodal data for cross-platform consistency • Maintained >98% quality score via self-audit and peer review

2021 - 2021

Education