Aircraft Appearance Assessment Tool
Developed a machine learning web app for the United Airlines QA team, leveraging TensorFlow to automate image classification from Gmail and Twitter — reducing manual QA time from hours to seconds.
Built a dashboard with 3 custom CNN models (ResNet-50 with image augmentation) to classify images as garbage, interior, or exterior and assess aircraft appearance scores. Designed ETL/ELT pipelines using Python and the Tweepy API to feed data into the models.
Deployed a React frontend with a Flask/Firebase backend for scalable architecture. Containerized with Docker in an Agile Scrum environment.
github.com/anthonykovari/CSE498-UAQA-GHUB ↗Dashboard
