Machine Learning Senior Capstone — United Airlines
Built a machine learning web app that scored plane interiors/exteriors on wear & tear and informed repair teams when defects were found, reducing manual QA time from hours to seconds.
Utilized TensorFlow, ResNet-50, and image augmentation to build three custom convolutional neural networks that powered the aircraft image classifier. Created the UI using React, built the backend with Python/Flask/Firebase, and containerized it with Docker.
Built an ETL/ELT data pipeline to ingest images from emails and tweets for classification using the Python Twitter and Gmail APIs.
github.com/anthonykovari/CSE498-UAQA-GHUB ↗Dashboard
