At AcSYS Engineering, I developed Python image-processing pipelines for visual inspection and classification, using OpenCV and PyTorch. I also adapted pretrained RF-DETR and Hugging Face vision models for object detection and segmentation workflows.
I built an SVM-based color classifier using HSV-derived image features to distinguish material colors.
At Northrop Grumman, I designed a Mixture-of-Experts architecture for impact-point prediction that improved generalization across 10 missile models. I also improved inference reliability by 30% through simulation-based evaluation of threat-prediction workflows.
My project work includes developing and integrating a C/C++ module for VLC Media Player to decode and play DVD-Audio formats, and proposing a probabilistic ML pipeline to identify airborne microplastic concentrations from satellite imagery.
