Liverson Ngugi

I’m a detail-oriented and resilient Data Annotator, Content Moderator, and AI Assistant Builder with a strong technical and analytical foundation. I develop tailored AI assets by combining data integrity, strict policy enforcement, and careful quality assurance to produce high-accuracy training data for advanced machine learning models.

As an AI Assistant Builder at The Co-operative University of Kenya, I engineered and optimized custom AI assistant frameworks, sourcing, cleaning, and structuring localized datasets to improve model response accuracy. I conduct quality assurance (QA) testing on model outputs, identify algorithmic biases, and refine prompt structures to ensure safety and alignment, including edge-case identification for robust fallback logic.

Freelance, I work as an independent Data Annotator and Content Moderator for Toloka AI, evaluating and filtering large volumes of user-generated content—flagging policy violations, adult content (NSFW), hate speech, and graphic imagery. I consistently achieve a 95%+ accuracy score by maintaining absolute neutrality and mental resilience, ensuring zero personal bias while handling highly explicit and sensitive datasets with absolute confidentiality. I’m also building on my foundation through a BSc in Accounting at The Co-operative University of Kenya, strengthening my quantitative methods and statistical analysis approach.