Udoh joy

I’m an AIData Specialist with 2+ years training multimodal and text-based LLMs through data annotation, SFT, RLHF, and evaluation tasks. I’ve completed 1,000+ projects across Tarmack, Turing, DataAnnotationTech, and Remotasks with 98%+ quality approval.

I specialize in both video data collection using mobile ultra-wide cameras and text-based reasoning tasks. I craft fact-checked text responses to train LLM reasoning and nuance, and I generate edge cases to improve performance and diversity.

I evaluate 1,500+ AI outputs for helpfulness, harmlessness, truthfulness, and style compliance, rewriting to meet C2 English standards. I also author detailed JSON metadata for videos/images and collaborate with QA teams, implementing feedback within 2 hours and improving team quality outcomes.

I’ve authored Nigerian/West African cultural edge cases and documented solutions to reduce bias for African users. I’m currently looking for remote, project-based roles in AI model training, evaluation, and writing where detailed guidelines and fast iteration matter.