Abbas Kazemipour is an applied scientist in Amazon’s search relevance team where he works on feature design, optimization and modeling to improve search. Prior to joining Amazon, he built deep learning models for anomaly detection in heart data (ECG), developed methods for data acquisition and analysis in neuroscience (e.g. compressed sensing application to microscopy, polynomial PCA for nonlinear dimensionality reduction), as well as towards more formal theory in deep learning (e.g. loss landscapes of deep quadratic networks).
