ONESMUS KITHOME

I am a detail-oriented AI Data Specialist with experience in AI training, prompt engineering, and written content evaluation. I craft high-quality, policy-aligned responses for AI training datasets and apply Reinforcement Learning from Human Feedback (RLHF) frameworks.

In my current roles, I write Prompt–Response–Preference (PRP) pairs to train and fine-tune LLMs, and I evaluate AI-generated responses for accuracy, tone, safety, helpfulness, and policy compliance. I provide structured feedback to improve model performance by ranking and comparing model outputs.

I also support consistent quality expectations across remote, asynchronous projects—maintaining steady output quality scores and applying model evaluation, data annotation, and prompt engineering practices. I collaborate with AI teams and the broader AI community to share best practices, raise quality flags, and iterate on guidelines.

I bring strong written communication and judgment in evaluating model output for safety, coherence, and alignment with complex policy requirements. My professional ethos is committed to responsible, ethical, and safe AI development.