Thandeka Ngubeni

I am an AI specialist with hands-on experience creating and curating high-quality linguistic datasets in English and Zulu. I focus on producing culturally relevant, guideline-compliant content that improves AI natural language understanding and model safety.

My recent work includes authoring over 2,000 original Zulu sentences for textual construction projects and serving as a Quality Assessor for Zulu content, demonstrating consistent accuracy and attention to linguistic nuance.

I have contributed to content moderation and safe-completion efforts by transforming AI-generated toxic prompts into benign, culturally appropriate outputs, and I have annotated multimodal data (video, text, images) to support computer vision model development.

I bring strong attention to detail, localization and transcription skills, and a commitment to ethical AI practices, seeking roles where I can apply my linguistic expertise to improve model performance and cultural alignment.