Amirhesam is a first year PhD student at UCSD advised by Misha Belkin. His research aims to understand i) when do neural networks work/fail, ii) interpolation v/s over-fitting, iii) whether kernel methods can be a serious rival to deep learning.