Voncarlos Marcelo de Araújo

At WELL-E AI Serving Welfare, I built and deployed a real-time cattle-vision pipeline that turns 320+ hours of field video into persistent identities, trajectories, and behavioral data for herd welfare analysis. I owned the full AI/ML lifecycle, from architecture and experiments through CUDA/TensorRT inference, deployment, monitoring, and production decisions.

I've also developed whale and beluga photo-identification systems, evaluated adversarial ML risks across 120K+ samples, and built multi-view plant-recognition models. My work combines computer vision, machine learning engineering, robust evaluation, and practical collaboration with animal-science teams.