About

A. Pasha Hosseinbor is a highly skilled Machine Learning Scientist at Amazon with over 9 years of relevant experience. He is an engineer by profession and an applied mathematician by heart. His specialties include Artificial Intelligence, Science, Programming, and Research. Pasha has a deep understanding of algorithms, math, statistics, physics, computer vision, estimation theory, and information theory. He is proficient in programming languages such as C/C++, Python (PyTorch, TensorFlow, Pandas, Scikit-learn & NLTK), R, and MATLAB. Currently, Pasha works at Amazon as a Machine Learning Scientist, where he focuses on improving personalization and diversification for customers in the clothing space. He owns a gamified ratings user experience product that can be thought of as the "Tinder for clothing". Pasha's work is based on transfer learning for fine-tuning, feature-extraction, or zero-shot scoring. He has developed a model that ranks items by how fashion-forward they are, and incorporated its signal into their ML recommender system. Additionally, he launched a model to filter out inappropriate items from their catalog, which saved 370 stylist hours per season and increased their catalog size by 50.2%. Prior to joining Amazon, Pasha worked at AMEX as a Data Scientist, where he developed a text clustering algorithm, an anomaly detection algorithm, and an NLP-based self-email identification algorithm. He also developed two different NLP-based algorithms for de-noising of email data and an NLP-based algorithm that matches security terms to their corresponding definitions using information retrieval. He is currently developing a predictive model for the internal security group at AMEX that predicts whether a given self-email is worthy of investigation or of no concern. As a Senior Machine Learning Scientist at Bio-Key International, Inc., Pasha developed a series of fingerprint recognition algorithms. He applied machine learning to the problem of pattern recognition in fingerprint feature space, with applications to image registration and comparison. He developed a machine learning framework to determine whether features from two different images match and implemented an algorithm for feature extraction from binary skeletonized fingerprint images. Pasha holds a Ph.D. in Medical Physics from the University of Wisconsin-Madison and a Bachelor's degree in Physics from the University of Virginia. His tech stack includes Research Scientist, ML, ResNet, NLP, Python, BERT, SVM, Scikit, Computer Vision, Image Processing, Deep Learning, Pandas, SQL, Information Retrieval