During my tenure as a post-doctoral fellow, I had the opportunity to work with many excellent faculty members, graduate students, and fellow postdocs that have helped me to launch my academic career. Intellectual Merit:As an NSF funded postdoc, I worked on a number of intellectually stimulating projects and papers, including:- Fast splitting methods for large-scale optimization and model fitting: This projects was focused on the development and theoretical analysis of accelerating the convergence of low-complexity methods for large-scale model fitting and distributed computing. I proved the strongest available theoretical convergence guarantees for accelerated first-order methods for constrained problems. I also created an easy-to-use implementation of my methods.- High-order schemes for sparse model fitting: First-order schemes can quickly solve sparse model fitting problems that are fundamental to machine learning and high-order statistics. My work in this area was focused on using high-order information (second derivatives) to solve problems much more rapidly than conventional methods. Broader Impacts: All of my work as an NSF post-doc was published in open access journals, and posted online in technical report archives for broad dissemination. I also produce open-source code repositories for all of my work. This enables people who are not experts in numerical analysis, including users of machine learning and data analysis, to use these tools to solve their problems efficiently.
