Jeffery (Junyan) Wu

Led development of smartphone-based skin condition screening systems using ensembling deep learning models on TensorFlow and AWS. Developed facial condition detection and acne severity grading models using multitask CNNs. Deployed and managed AI models in cloud environments using Docker and TensorFlow Serving. Collaborated with clinical researchers to support academic publications and FDA submissions.

Led product-driven development of multiple AI models and pipelines for clients including Walmart, Two Sigma, Merck, and Albertson. Developed security models for abnormal user behavior detection, deep learning models for smart inventory management, semi-supervised pipelines for plant phenotyping, and OCR pipelines for railroad images. Migrated databases and data pipeline systems across major cloud platforms. Built marketing recommendation systems and MLOps pipelines leveraging AWS technologies. Also worked on real-time drone video captioning using advanced video-language models.

Joined as first employee and developed FDA 510k cleared whole heart segmentation U-Net model on PyTorch. Created vessel smart-skeleton algorithms innovating snake algorithms. Designed and implemented clinical tests for regulatory qualification. Worked on vessel pathfinding algorithms and automatic CT scan quality ranking using novel CNN training techniques. Developed distributed ML prediction API services on AWS.