Assistant Professor at Northeastern University.My lab studies the emergent internal mechanisms of deep networks.Characterizing the structure of learned computation is a grand and fundamental research problem with many practical implications that will be the foundation for a new generation of computer science. My lab publishes fundamental research in Machine Learning, Computer Vision and Natural Language Processing.MIT PhD dissertation: Dissection of Deep Networks. First-author peer-reviewed research published in PNAS, CVPR, SIGGRAPH, ECCV, ICLR, ICCV, and ACM Communications. Creator of GAN Paint, Network Dissection, and the ROME method for editing knowledge inside GPT.Previously, new product development at Google. Helped develop Google Image Search, Google Talk, Apache XML Beans, BEA Weblogic Workshop, Microsoft Internet Explorer, the .NET Regular Expression Compiler, "Dave's Quick Search Deskbar," and the Pencil Code educational programming system.Co-author of "Numerical Linear Algebra", a widely-used textbook on numerical methods.Specialties: Novel deep network methods, Interactive visualizations and interfaces, Numerical methods.