Led a development team using Python to build a machine learning model using Mask - RCNN classification that would recognize and then classify damaged and undamaged cars based on user-provided images.
Partnered with a developer to implement a RESTful API in Flask to enable users to upload images.
Used Scikit-learn and Scikit-image, for machine learning, and image processing respectively.
Drove a unit test effort to improve the accuracy of the model from 56% to 90%.
Led process to deploy the application onto Azure. Used Azure Cognitive Services for the ML models and Azure Custom Vision for improving the accuracy of processing images by up to 99.9%.