Marc Roy Email & Phone Number

-Built full-stack machine learning app for determining if a supplier issue is systemic in a historical record of free-text based documents that track supplier problems. Accomplished this by implementing an embedding model and vector database to determine if similar documents/issues have been seen before and how often that problem has occurred.-Deployment.

United States

- Queried, cleaned, and transformed 450 million time series IIOT data points.- Engineered and tested features by feeding them into various models (PCA, Random Forest, LSTM) and identifying anomalies for predictive maintenance.-Created a KPI for Industrial CNC machine productivity and progress using IIOT data and Grafana. Deployed on factory floor for.

Boulder, Colorado, United States

– Built birds eye view car simulation using Python/pygame/numpy. Car was approximated using a simple kinematic bicycle model and collision detection and reference lanes were built with pygame. – Integrated the simulation with a Deep Q-learning neural network with custom environment and reward system for the purpose of navigating paths. Tested different.

Boulder, Colorado, United States

-Helped process camera footage using Python and the OpenCV library. Rewrote functions to accept videos with AVI format to cut out AVI to MP4 conversion and reduce processing time.-Used Python and OpenCV to run a set of difficult to detect car/pet/pedestrian test images through YOLOv3 on the Darknet framework. This helped to identify problems in autonomous.

Boulder, Colorado, United States

-Independently built game to be used by both human participants and AI to compare their differences in movement and intent.-Used C# and Unity to build necessary game scripts and assets.