Kelvin Njenga

I’m an AI Trainer and Data Labelling Specialist focused on producing high-accuracy labeled data that directly improves machine learning model performance. I bring a hands-on, detail-first approach to data annotation, quality control, and AI training methodologies across computer vision and natural language processing.

Since January 2023, I’ve annotated 15,000+ image assets for object detection models while maintaining a 98.5% inter-annotator agreement score. I also developed annotation guidelines for ambiguous edge cases, reducing reviewer escalations by 40%, and worked with ML engineers to refine the labelling taxonomy using feedback loops.

Previously, I served as a Content Reviewer & Data validator, evaluating user-generated content against safety policies with 99.2% accuracy. I flagged nuanced policy violations that required human judgment, documented labeling inconsistencies, and trained five new team members on annotation tools and quality standards.

I’ve applied my skills through projects like Autonomous Vehicle Perception—where I led segmentation labeling for 2,000+ street scene images—and Multilingual Text Classification, categorizing 10,000+ support tickets across 3 languages with consistent taxonomy. I’m committed to responsible, well-documented processes, supported by my BSc in Data Science and training in AI ethics and responsible data labelling.