Thomas Patton Email & Phone Number

San Diego, California, United States

- Created a hardware-optimized computer vision library in C/C++ designed to run on lightweight drones. This library offloads core computer vision logic to signal processors, resulting in a significant reduction in CPU utilization, power consumption, and heat generation.- Used aforementioned library to create a custom visual feature tracker in C++ to assist in Visual-Inertial Odometry. As a result of hardware optimizations, tracker is significantly more efficient enabling performant real-time tracking on-device.- Assumed ownership and maintenance responsibilities for key packages within the company's SDK focused on computer vision including our machine learning models, depth from stereo server, and thermal optic flow.

Ithaca, New York, United States

- Worked in Cornell University’s EmPRISE lab to engineer a caregiving robot to assist with feeding. Project involves developing Deep Learning models in PyTorch to expand the number of recognizable food items. Integrated model using computer vision, Python, C++ and ROS. Advised by Tapomayukh Bhattacharjee.

Boston, Massachusetts, United States

- Designed a scalable and transparent Optical Character Recognition pipeline to process uploaded clinical documents using Tesseract, AWS, Docker and Kafka. Pipeline operates 85% faster than its predecessor leading to a large operational savings.- Developed Natural Language Processing pipelines using SpaCy to extract clinical information from a patient’s documents. These pipelines assist patients in being approved for insurance creating cost savings while reducing burden on nurses.- Worked in an agile-based team using Git, AWS, and Docker.

- Developed quantitative lung airway features to be used in a machine learning model to predict patient response to immunotherapy treatment leading to a classifier able to predict with an AUC of 0.67. Second-authored a publication on the project: https://jitc.bmj.com/content/9/Suppl_2/A44- Implemented a post-processing algorithm which takes CT scans whose lung boundaries have been corrupted by COVID-19 and automatically corrects them. This algorithm is part of a fully-functional pipeline using Python and OpenCV which automates the lung segmentation imaging process for radiologists and is under peer-review for publication in a medical journal.

Cleveland/Akron, Ohio Area

- Created an algorithm using C and OpenGL which takes spaceflight parameters for the International Space Station and generates a visualization indicating the spacecraft’s orientation in flight. This provides insight to how self-shadowing occurs on the solar panels during orbit - an invaluable resource to the power analysis team at NASA to assist with solar panel and battery development.

Master Of Engineering - Meng, Computer Science

Bachelor Of Science - Bs, Computer Science

High School Honors Diploma, High School/Secondary Diplomas And Certificates