Jacob Feitelberg

Jacob Feitelberg

jef2182@columbia.edu

Columbia University
Industrial Engineering and Operations Research

I am a Ph.D. candidate in Operations Research at Columbia University. I am currently being advised by Prof. Anish Agarwal and working on projects on tabular foundation models, causal inference, and artificial intelligence. I completed my M.S.E. in Applied Mathematics and Statistics at Johns Hopkins University in 2023. My thesis, titled “Globally Optimal Matching of Astronomy Catalogs Using Mixed Integer Quadratically Constrained Programming and Constrained Clustering,” was advised by Prof. Amitabh Basu and Prof. Tamás Budavári.

My interests include tabular representation learning, optimal transport, causal inference, and optimization.

Projects

MatrixFormer: a foundation model for matrix completion. A pretrained transformer fills in the missing entries of any numerical matrix, zero-shot, with no training or fine-tuning.
Project page · Code · Paper

TabImpute: zero-shot transformer imputers for tabular missing data. One pre-training pipeline builds a robust default model and pattern-specific specialists, evaluated on the MissBench benchmark.
Project page · Code · Paper

Publications

Preprints

MatrixFormer: A Foundation Model for Matrix Completion Dwaipayan Saha*, Jacob Feitelberg*, Kyuseong Choi*, Raaz Dwivedi, Anish Agarwal

Project page · Code · Weights · *Equal contribution

One Pipeline, Many Transformers: Pattern-Specific Imputation Specialists for Tabular Missing Data Jacob Feitelberg, Dwaipayan Saha, Kyuseong Choi, Zaid Ahmad, Anish Agarwal, Raaz Dwivedi

Project page · Code · Weights
Spotlight at the Causal Inference Workshop @ ACM SIGMETRICS 2026.

N²: A Unified Python Package and Test Bench for Nearest Neighbor-Based Matrix Completion Caleb Chin, Aashish Khubchandani, Harshvardhan Maskara, Kyuseong Choi, Jacob Feitelberg, Albert Gong, Manit Paul, Tathagata Sadhukhan, Anish Agarwal, Raaz Dwivedi

Accepted to CODEML Workshop @ ICML 2025.

Distributional Matrix Completion via Nearest Neighbors in the Wasserstein Space Jacob Feitelberg, Kyuseong Choi, Anish Agarwal, Raaz Dwivedi

Accepted to Causal Inference Workshop @ ACM SIGMETRICS 2025.

Learning Counterfactual Distributions via Kernel Nearest Neighbors Kyuseong Choi, Jacob Feitelberg, Anish Agarwal, Raaz Dwivedi

Journal Articles

Fast Catalog Matching for Improved Posteriors via Constrained Clustering Jacob Feitelberg, Amitabh Basu, and Tamás Budavári
Fast Globally Optimal Catalog Matching using MIQCP Jacob Feitelberg, Amitabh Basu, and Tamás Budavári
MNE-ICALabel: Automatically annotating ICA components with ICLabel in Python Adam Li, Jacob Feitelberg, Anand Prakash Saini, Richard Höchenberger, Mathieu Scheltienne

Theses

Master's Thesis: Globally Optimal Matching of Astronomy Catalogs Using Mixed Integer Quadratically Constrained Programming and Constrained Clustering Advisors: Amitabh Basu and Tamás Budavári