Amin Ullah

■ Innovative Machine/Deep Learning expert with in-depth expertise in Computer Vision, including Human action/activity recognition, anomaly detection, image/video retrieval, and person re-identification, with a passion for research and development in the field.
■ Led and managed projects leveraging expertise in team management, knowledge transfer, and developing algorithms for image/video analytics, resulting in 40+ peer-reviewed publications and 2 book chapters.
■ Strong communication skills, ability to collaborate effectively, manage dedicated projects and co- authored 18 publications and 10 successful collaborations with global scientists and developers.

Postdoctoral Scholar Jun 2021 - Present Oregon State University, Oregon, USA
■ Conducting research on novel object/action detection in open-world scenarios to develop scientific principles for quantifying/characterizing novelty and creating AI systems that can handle it effectively.
■ Exploring deep generative models for optimized neural network generation, with the goal of developing a model that can efficiently sample trained networks.
■ Investigating uncertainty in underwater object analysis, developed a dataset for evaluating uncertainty and currently conducting experiments to improve performance through analyzing uncertainty in object segmentation.
Postdoctoral Scholar Jan 2021 – May 2021 Sejong University, Seoul, South Korea
■ Managed projects and priorities, coordinating 8 researchers leading to the timely completion of 3 projects on video data analysis technology for intelligent surveillance systems.
■ Published a survey on the reliability and efficiency of deep architectures for real-time performance and overviewed state-of-the-art strategies for safe autonomous driving.
■ Developed 2 computationally efficient networks for person re-identification in surveillance videos, examining autoencoders, channel attention, spatial attention, and dilated convolutional networks.

Graduate Research Assistant Mar 2017 – Jan 2021 Sejong University, Seoul, South Korea
■ Developed deep learning techniques for human action recognition in videos using CNNs, RNNs, LSTMs, resulting in 6 leading author papers, ranked top 0.1% most cited articles published in 2017, 2018, and 2019 in the computer science category.
■ Assisted in developing algorithms for video surveillance and movie summarization using target- appearance-based features, emotional moments and facial expression recognition, resulting in 20 publications.
■ Created a real-time baby behavior monitoring system using a Raspberry Pi device with a vision sensor to detect anomalies and generate alerts for parents and caregivers.
Deep Learning Developer Jan 2018 – Dec 2018 Seoho Electric Co., Ltd, South Korea
■ Developed real-time container corner cost detection system for automatic ships loading at ports, utilizing deep learning technologies in Python, C# for the front-end, and C++ for real-time video decoding for efficient performance.

Ph.D., Engineering in Digital Contents, Sejong University Mar 2017 – Feb 2021
Thesis: A Study of Sequential Patterns Analysis in Video for Action and Activity Recognition using Deep Learning.
BS Computer Science, Islamia College Peshawar, Pakistan Sep 2012 – Oct 2016
CGPA (3.93/4.0) and Awarded Gold Medalist in BS-Computer Science.