Alexander R. Loftus

Alex Loftus is a Kaggle competition winner, and PhD student under advisor David Bau. His research focuses on interpretability in code models and training dynamics. His background includes roles in data science, machine learning engineering, and biomedical machine learning at Johns Hopkins University. He helped organize the New England Mechanistic Interpretability conference, contributed to work on interpretability infrastructure, and was part of the first-place Vesuvius Ink Detection Kaggle team featured in Scientific American. He also created a linear algebra tutorial series building up to the mathematical foundations of neural networks. His interests span spectral theory, information geometry, the history of science and mathematics, and ethics.