Research
My work sits between optimization and applied modeling, on problems where a decision follows from a computation: what a property is assessed at, what a facility draws from the systems around it, what a solver certifies as optimal. In each of these the interesting question is not only accuracy but structure — which errors are acceptable, which are systematic, and what the method can honestly claim.
Three themes organize the work. They share a discipline: make the decision-relevant structure explicit, distinguish what is observed from what is modeled, encode the important trade-offs when the decision itself is optimized, and evaluate under the conditions the application will actually face.
Predictive systems whose errors carry distributional or institutional consequences — where accuracy alone is not the right objective.
Energy, water, and environmental systems reconstructed from imperfect public records, with an explicit boundary between reported and modeled quantities.
Decomposition, sparsity, and first-order methods on modern hardware, with numerical reliability that survives scale.
Work I am doing now at the MIT Operations Research Center.
Fair & reliable predictive modeling
A mass-appraisal model can be made more accurate and still leave some owners systematically over-assessed; the two goals have to be optimized together.
Infrastructure modeling & planning
What a large data center draws from the grid, the water system and the aquifer around it can be partially reconstructed from public records — provided the reconstruction is explicit about which quantities are reported and which are modeled.
Scalable optimization & computation
A first-order solver that runs fast on a GPU is only useful if its termination criteria still mean what they meant on a CPU.
Completed work that shaped the methods above.
Scalable optimization & computation
Master's research on a scalable decomposition approach for feature selection under many correlated candidate predictors.
Fair & reliable predictive modeling
An interpretable anomaly-screening protocol for Chile's university selection exam.
Fair & reliable predictive modeling
Online optimization for adaptive questionnaires, formulated as a variant of matrix completion.