Paul Grigas

Timeline

1. 2011

   Operations research at Cornell

   He earns a BS in Operations Research and Information Engineering at Cornell University in 2011.

2. 2015

   A prize for boosting

   In 2015, he wins the INFORMS Optimization Society Student Paper Prize for “A new perspective on boosting in linear regression via subgradient optimization and relatives,” coauthored with Robert M. Freund and Rahul Mazumder. The paper analyzes boosting algorithms for linear regression through first-order methods in convex optimization.

3. 2016

   MIT doctorate and Berkeley faculty

   He completes a PhD in operations research at MIT’s Operations Research Center in the Sloan School of Management in 2016, with Robert M. Freund as his advisor. His thesis, “Methods for convex optimization and statistical learning,” studies algorithms for convex optimization and statistical machine learning. In 2016, he joins UC Berkeley’s Department of Industrial Engineering and Operations Research as an assistant professor.

4. 2017

   Smart predict, then optimize

   With Adam N. Elmachtoub, he releases the preprint “Smart Predict, then Optimize” in 2017. Their framework trains prediction models using the structure of the optimization problem that uses the predictions; its SPO loss measures the decision error caused by a prediction, and its SPO+ loss offers a tractable training objective.

5. 2018

   An NSF grant for optimization

   In 2018, he receives a $175,000 NSF Computer and Information Science and Engineering Research Initiation Initiative grant to develop optimization algorithms for large datasets. The project studies the Frank-Wolfe method and in-face directions to improve memory use and preserve useful model structure.

6. 2020

   First place for SPO

   He wins first place in the 2020 INFORMS Junior Faculty Interest Group Paper Competition for “Smart Predict, then Optimize,” coauthored with Columbia University’s Adam Elmachtoub.

7. 2021

   Online learning and decisions

   In 2021, he coauthors “Joint Online Learning and Decision-making via Dual Mirror Descent” with Alfonso Lobos and Zheng Wen at the International Conference on Machine Learning. The work combines a method for learning unknown parameters with online decisions to maximize revenue under cost constraints.

8. 2021

   Integrated conditional estimation and optimization

   With Meng Qi and Zuo-Jun Max Shen, he releases “Integrated Conditional Estimation-Optimization” in 2021. Their framework estimates conditional distributions of uncertain parameters using an objective aligned with the optimization problem that will use those estimates.

9. 2025

   Learning the optimal solution path

   He coauthors “Beyond Discretization: Learning the Optimal Solution Path” with Qiran Dong and Vishal Gupta at the International Conference on Artificial Intelligence and Statistics in 2025. The method represents a solution path with basis functions and learns the path through a single stochastic optimization problem.

10. As of October 2026

    Associate professor at Berkeley

    As of October 2026, he is an associate professor in UC Berkeley’s Department of Industrial Engineering and Operations Research and an affiliate of the NSF AI Research Institute for Advances in Optimization (AI4OPT). He develops machine learning models for decision-making and designs and analyzes optimization algorithms for machine learning and operations research.