Connor Schmidt

Dedicated product support specialist with a background in end-user software validation testing, creating technical and support documents, and providing software integration and testing support for customers. A diligent, communicative, and team-oriented worker seeking to leverage my experience and education in Computer Science and Machine Learning. Motivated to take on responsibility and apply myself to projects in Artificial Intelligence and Software as a Service industries.

Machine Learning Attack and Defense: Python, PyTorch

• Collaborated in a team environment implementing various attack and defense methods against a trained image classification model and conducted research of different technical publications and diagrams to implement algorithms which improved the attack methods.
• Developed implementations of a separate attack detector classifier, image preprocessing using Gaussian blur and randomized crops/flips, and adversarial training to defend an image classification model against FGSM and PGD attacks.
• Developed implementations of a modified FGSM and PGD attack approaches utilizing the classification model, detector model, and a dynamic step size growth/decay to attack an image classification model.
• Created, updated and delivered documentation and reports throughout development cycle on schedule.

Ensemble of Classifiers: Python, PyTorch

• Applied and developed organizational and planning skills in a team environment to divide labor and meet implementation goals and deliver reports on schedule.
• Supported planning, scheduling, and development of a machine learning classifier model utilizing a weighted bagged ensemble and satellite data of cloud temperatures to predict rainfall at certain geographic coordinates.
• Supported writing and delivery of team reports on progress and changes made throughout the development cycle, delivered capabilities, and final analysis and conclusions on the project.