Summary
Harry Wang is an applied scientist in machine learning with nine years of experience building production ML and deep learning systems across recommendation, ads ranking, NLP, and graph learning. Currently at Amazon after driving CTR/CVR and inference-cost improvements at Pinterest, he blends research-grade modeling (BERT, transformers, contrastive losses, knowledge distillation) with pragmatic engineering (TensorFlow→PyTorch migrations, mixed-precision inference, GPU serving). His academic work at Penn and Michigan produced published research on misinformation detection, stochastic optimization, and graph mining, and he has taught graduate deep learning and AI courses to large cohorts. Comfortable spanning the full ML lifecycle, Harry has shipped multilingual NLU pipelines, large-scale PySpark feature engineering, and production-grade ensemble models. He’s equally at home prototyping novel architectures and optimizing operational cost, and often focuses on stabilizing models in nonstationary (daily-rebasing) environments.
9 years of coding experience
6 years of employment as a software developer
University of Michigan College of Engineering
Penn Engineering
Bachelors of Science in Engineering (with Honors) - BSE Computer Science (Major) Mathematics (Minor), Bachelors of Science in Engineering (with Honors) - BSE Computer Science (Major) Mathematics (Minor) at University of Michigan
High School Diploma Science Technology Engineering Mathematics (STEM), High School Diploma Science Technology Engineering Mathematics (STEM) at The Academy for Mathematics, Science, and Engineering
Masters of Science in Engineering - MSE Computer Science (CIS) | Artificial Intelligence Concentration, Masters of Science in Engineering - MSE Computer Science (CIS) | Artificial Intelligence Concentration at University of Pennsylvania
French, English, Chinese