dennis rubui

At Outlier, I evaluate AI-generated image and video outputs using detailed quality rubrics, delivering preference judgments that help fine-tune generative models. I assess human-activity detection accuracy by verifying true and false positives and negatives, annotating edge cases, and reporting issues that improve model reliability. I also cross-check factual claims against source URLs to assess citation reliability and correctness for research-focused training data. I built a real-time Power Line Fault Detection & Alert System using anomaly-detection algorithms, Firebase, and a live dashboard. I independently managed testing, iteration, and accuracy tuning. My electrical and electronics engineering background supports a systematic approach to documentation, testing, data logging, and analytical problem-solving.