I am in the fifth and final year of my PhD at MIT Department of Mathematics, co-advised by Ankur Moitra and Jon Kelner, and working at the interface of data science, machine learning, optimization, statistical physics, and probability.

I design and study algorithms to analyze very noisy data. Much of my work is motivated by estimating 3D molecule structures from cryo-EM images, and I study families of related noisy geometric problems that apply to robotics, image processing, signals processing, community detection in networks, and more. My work draws from random matrix theory and other theory of random structures; from statistical physics, the replica method, and the idea of phase transitions in data; from representation theory and invariant theory to exploit rich problem geometry and symmetry; and from convex optimization, semidefinite programming, and the sum of squares hierarchy as tools and as a perspective on algorithms in general.

I spent my undergraduate years at Keble College, Oxford, mainly thinking about algebraic topology. Code for various topology computations is available on my github page.


Send emails to amelia at this domain, or to ameliaperry at mit.
My office at MIT is 2-390D. map