Summary
Zarrar Shehzad is a Data Scientist III with 16 years of experience applying machine learning, NLP, and Bayesian deep learning to product-grade problems across audio, video, and human behavior. At Audible he’s improved search relevance with bi-encoder and XGBoost models, built multilingual LLM moderation systems, and architected an end-to-end conversational search explanations pipeline on SageMaker. Previously at CLIPr he drove a tenfold cost reduction via automated video annotation and pioneered speaker identification using Bayesian deep learning. His background as a cognitive neuroscience PhD and postdoc informs a strong grounding in experimental design and interpretable modeling, evidenced by grant-funded work and datasets used by other researchers. Based in the NYC area, he blends research rigor with pragmatic engineering to ship scalable AI solutions that measurably boost user engagement. An understated strength is his track record of translating complex cognitive models into production-ready pipelines that cut costs and improve product relevance.
16 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Cognitive Neuroscience (Psychology), Doctor of Philosophy - PhD, Cognitive Neuroscience (Psychology) at Yale University
Bachelor of Science - BS, Psychology, Bachelor of Science - BS, Psychology at McGill University