Wesley Ekpenyong

Evaluated AI-generated responses for accuracy, relevance, and safety, reviewing hundreds of outputs weekly to ensure compliance with ethical AI standards. Flagged hallucinations, biases, and logical errors by cross-referencing sources and guidelines, then provided detailed corrective feedback for model refinement.

Completed short-term chatbot testing, image sorting, and AI training data collection tasks requiring precise labeling. Built and validated datasets for conversational AI and computer vision by categorizing images for object detection and testing dialogue flows for logical consistency, using feedback loops to improve dataset quality and reduce hallucinations.

Enhanced large-scale machine learning models through high-quality data labeling across natural language processing, image recognition, and predictive analytics. Processed thousands of data points to create clean training sets, identified inconsistencies using statistical modeling and cloud-based tooling, and improved precision and reduced error rates via iterative guideline reviews.

Annotated data for search relevance, content moderation, and multimedia tasks across text, images, and videos to enhance algorithm accuracy for global clients. Conducted sentiment/relevance/compliance labeling, performed quality control with inter-annotator agreement, and helped improve search results and safer platforms through culturally sensitive guidelines.

Performed flexible annotation and data collection for text, images, and audio datasets supporting AI training in healthcare and multilingual content. Identified inconsistencies in medical imagery and text to reduce biases, then collaborated to refine annotation protocols and improve data integrity and labeling efficiency.

Earned a Bachelor of Science degree at the University of Uyo in Uyo, Nigeria, graduating in 2023.