Stephanie O.

Key Skills

Software

Appen

Top Subject Matter

Nigerian Pidgin NLP

culturally specific emotion and sarcasm annotation taxonomy

Nigerian market intelligence chatbot

Top Data Types

Text

Document

Top Task Types

Emotion Recognition

Prompt + Response Writing (SFT)

RLHF

Entity (NER) Classification

Classification

Freelancer Overview

Founder & Lead Researcher, Wazobia Labs (Wazobia Eval). Brings 3+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include HuggingFace, GitHub, and Google Sheets. Education includes Bachelor of Science, Federal University of Technology, Owerri (FUTO) (2023). AI-training focus includes data types such as Text and labeling workflows including Emotion Recognition, Prompt + Response Writing (SFT), and RLHF.

Labeling Experience

Founder and Lead Researcher - Wazobia Labs

AppenTextRLHFEntity (NER) Classification

Founder and lead researcher at Wazobia Labs building Nigerian Pidgin emotion and culture evaluation resources for NLP research. She developed benchmark design, annotation schema, and evaluation methodology grounded in culturally specific registers and meaning. The work required strong research thinking, codebook and taxonomy development, and end-to-end dataset publishing and quality control. • Built and published Wazobia Eval benchmark for Nigerian Pidgin emotion understanding, sarcasm detection, and cultural reasoning • Developed a 16-category Nigerian emotion taxonomy capturing registers absent from existing NLP frameworks • Designed a production-grade annotation schema including dual emotion capture, prosody annotation, and tonal variant documentation • Managed end-to-end dataset pipeline and recruited annotators, including inter-annotator agreement workflows using Cohen's Kappa

2026 - Present

Founder & Lead Researcher, Wazobia Labs (Wazobia Eval)

TextEmotion Recognition

Founded and led end-to-end dataset creation for Wazobia Eval, focused on Nigerian Pidgin emotion understanding, sarcasm detection, and cultural reasoning. Developed a 16-category emotion taxonomy and a 15-column annotation schema including dual emotion capture, prosody annotation, sarcasm pairing, tonal variant documentation, and health domain coverage. Managed dataset pipeline from collection through quality control and HuggingFace publishing while maintaining the GitHub repository for the benchmark. • Designed codebook/taxonomy and standardized evaluation protocol • Implemented inter-annotator agreement workflow using Cohen's Kappa • Conducted pilot RLHF-style evaluation with GPT on culturally grounded examples • Oversaw 550+ manually annotated entries and label quality checks

2026 - Present

AI Product Builder (YouSabi) — Independent

TextPrompt + Response Writing (SFT)

Built culturally adapted AI chatbot and related AI products for Nigerian sales psychology using Anthropic API. Contributed to AI product development likely involving prompt writing and dataset preparation for desired conversational behavior. Focused on culturally grounded outputs rather than general-purpose content generation. • Designed chatbot behavior aligned to Nigerian buyer communication patterns • Supported AI-assisted development for functional, publicly accessible products • Authored prompts and response behavior guidelines for target audiences • Iterated product prompts based on user intent and cultural context

2025 - Present

Education

Work History