Interpretable Neural Networks Lab, advised by David Bau

• Computer Vision (CS 450) – Winter 2022 - We covered traditional computer vision methods including the hough transform, graph cut segmentation, image alignment and stitching, a project on cropping and aligning QR codes, and an intro of deep learning for vision.• Introduction to Deep Learning (CS 474) – Fall 2021 - We covered many deep learning fundamentals including CNNs, RNNs, transformers, GANs, loss functions, and RL; With applications such as a cancer detection U-Net, Gatys' original style transfer, text prediction, machine translation, and an introductory GAN.• Introduction to Machine Learning (CS 472) – Winter 2021, Summer 2021 - We covered traditional machine learning methods including perceptrons, multilayer perceptrons and backpropagation, decision trees, KNN's and clustering algorithms.

• Studied crowd dynamics & sentiment using both supervised and unsupervised machine learning methods on crowd noise datasets to discover and classify crowd states towards future intervention & controller design• Co-author of 2 academic publications (1 first author)• Developed and streamlined the data visualization process used to understand and interpret classification results• Worked on feature engineering & validation, and integrated audio feature processing into custom data pipeline• Presented research at BYU’s CPMS Student Research Conference (Mar 2018, Mar 2019)• Collected data at more than 25 BYU sporting events and helped build data labeling guidelines.

• Held weekly office hours for a Python-based class of 60+ students studying advanced mathematical analysis, algorithms, and optimization• Labs include various optimization methods, as well as algorithms like eigenfaces, KNN’s, iterative numerical solvers, Q-Learning, and policy iteration. I also taught data science tools such as regex, Unix shell, and SQL

• Expanded and improved a Python machine learning model to successfully predict academic failure for all BYU students in early weeks of the semester. The product is known as Early Alert and is used by most campus support offices to provide individualized support for students.• Reduced model training time 2x through code profiling and optimization• Daily tasks included data cleaning, data storage and retrieval using SQL, feature engineering, feature selection, algorithm optimization, and model evaluation.

• Assisted 20+ students in their second semester of Python to develop coding, proof, and problem-solving skills

• Linked 50,000+ vital genealogical records as part of the labeling process to train machine learning algorithms• Received the “Golden Widget Award” for exceptional labeling accuracy, work updates and analysis, and coworker collaboration