At YuVerse, part of Yubi Group, I build scalable AI video systems for personalized content production. I architected KEDA-driven autoscaling from 0 to 1,000 replicas, reduced GPU compute costs by 65–70%, and shipped a Real-ESRGAN face-enhancement pipeline that eliminated five recurring production artifacts.
I've built GPU-accelerated lip-sync services on AWS EKS and Temporal Cloud, a custom ffmpeg video-editing engine that cut processing time by about 75%, and automated evaluation pipelines for lip-sync drift and audio lag. My work also spans machine learning, data science, financial research, and dashboards that improved sales call success rates by 20%.
