Alfredo Torrico

I am a 5th-year Ph.D. candidate in Operations Research at the H. Milton Stewart School of Industrial and Systems Engineering, Georgia Tech.

I am currently in the job market, looking for postdoc and academic positions.

Contact

atorrico3 [at] gatech [dot] edu

Research experience and interests

My main research interest lies at the intersection of Combinatorial Optimization, Online Optimization, and Machine Learning. My previous and current work has focused primarily on new approaches to classical resource allocation and subset selection problems. I am very excited about these problems, especially because of their numerous applications in real-world instances such as online advertising, sharing-economy systems, feature selection, influence maximization, among others. However, there are many challenges that remain to be solved in the aforementioned areas, particularly those related to stability, information, efficiency and fairness. In the future, I would like to explore and expand my research scope to partial information models, and the design of efficient and fair algorithms.

During my first two years in the Ph.D. program, I worked with Alejandro Toriello and Shabbir Ahmed on a new polyhedral approach to the well-known online bipartite matching problem. After that, I have been working with Mohit Singh and Sebastian Pokutta on submodular optimization problems. In our first work, we study the offline and online versions of the robust submodular maximization problem under a wide class of structured combinatorial constraints. In a follow-up paper, we focus on efficient algorithms for the offline model, and we show that can be easily implemented in several applications. Currently, we are working on the notion of stability in submodular optimization.

I previously obtained a Mathematical Engineering degree (Applied Math M.Sc. equivalent) under supervision of Roberto Cominetti in the Mathematical Engineering Department at Universidad de Chile. During this period, we study a risk-averse axiomatic framework for routing problems.

For more details, see my research statement.