Abinitha Gourabathina

PhD Student · MIT EECS

Building methods for reliable and controllable ML models

About Me

Hello! I am a PhD student in MIT LIDS, co-advised by Professors Marzyeh Ghassemi and Collin Stultz. Prior to my PhD, I received my B.S.E. from Princeton University in 2023, where my senior thesis was supervised by the incredible Professor Christiane Fellbaum.

My research focuses on reliable and controllable machine learning with the goal of building models that behave predictably across diverse conditions. In particular, I find value in working on high-stakes domains like healthcare. I also enjoy drawing on tools from optimization, probability, and statistics in my work.

Outside of research, I love to read and have recently gotten into birding!

You can reach me at abinitha@mit.edu. I'd love to hear from you!

News

  • Apr 2026: Trace Inversion is accepted @ ACL Main Conference!
  • Apr 2026: Our paper Trace Inversion is up on arXiv!
  • Feb 2026: Our paper LEIA is released on arXiv!
  • Oct 2025: Presented an ML Tea Talk on trace inversion.
  • Sept 2025: Excited to serve as Program Chair of the 2026 LIDS Student Conference!
  • Aug 2025: Completed my summer internship at IBM Research!
  • June 2025: Our work is covered by MIT News and other press.
  • June 2025: Presented at FAccT '25!
  • June 2025: Paper released on arXiv!
  • May 2025: Excited to join the AIES PC!
  • May 2025: MedPerturb project website online!
  • April 2025: Presented @ the MIT EECS Town Hall for SERC.
  • April 2025: Paper accepted to FAccT '25!
  • April 2025: Presented @ the MIT IMES Seminar Series.
  • Aug 2024: Started my PhD at MIT!

Research Highlights

Answering the Wrong Question: Reasoning Trace Inversion for Abstention in LLMs

Trace Inversion

Abinitha Gourabathina, Inkit Padhi, Manish Nagireddy, Subhajit Chaudhury, and Prasanna Sattigeri, 2026. arXiv:2604.02230.

Robustness Beyond Known Groups with Low-rank Adaptation

LEIA

Abinitha Gourabathina, Hyewon Jeong, Teya Bergamaschi, Marzyeh Ghassemi, and Collin Stultz, 2026. arXiv preprint arXiv:2602.06924.

The MedPerturb Dataset: What Non-Content Perturbations Reveal About Human and Clinical LLM Decision Making

MedPerturb

Abinitha Gourabathina, Yuexing Hao, Walter Gerych, Marzyeh Ghassemi, 2025. arXiv:2506.17163.

The Medium is the Message: How Non-Clinical Information Shapes Clinical Decisions in LLMs

The Medium is the Message

Abinitha Gourabathina, Walter Gerych, Eileen Pan, and Marzyeh Ghassemi. 2025. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT '25).

PanDa Game: Optimized Privacy-Preserving Publishing of Individual-Level Pandemic Data Based on a Game Theoretic Model

PanDa Game

Abinitha Gourabathina, Zhiyu Wan, J. Thomas Brown, Chao Yan, Bradley A. Malin, 2023. In IEEE Transactions on NanoBioscience, vol. 22, no. 4, pp. 808-817.

What seems to be the problem? Stigmatizing language in patient medical notes

Stigmatizing language

Abinitha Gourabathina (Senior thesis, Princeton University). Princeton DataSpace. http://arks.princeton.edu/ark:/88435/dsp01cv43p110t

Papers

  • Answering the Wrong Question: Reasoning Trace Inversion for Abstention in LLMs
    Abinitha Gourabathina, Inkit Padhi, Manish Nagireddy, Subhajit Chaudhury, and Prasanna Sattigeri — ACL 2026
  • Robustness Beyond Known Groups with Low-rank Adaptation
    Abinitha Gourabathina, Hyewon Jeong, Teya Bergamaschi, Marzyeh Ghassemi, and Collin Stultz — Preprint, 2026
  • The MedPerturb Dataset: What Non-Content Perturbations Reveal About Human and Clinical LLM Decision Making
    Abinitha Gourabathina, Yuexing Hao, Walter Gerych, and Marzyeh Ghassemi — Preprint, 2025
  • The Medium is the Message: How Non-Clinical Information Shapes Clinical Decisions in LLMs
    Abinitha Gourabathina, Walter Gerych, Eileen Pan, and Marzyeh Ghassemi — FAccT 2025
  • PanDa Game: Optimized Privacy-Preserving Publishing of Individual-Level Pandemic Data Based on a Game Theoretic Model
    Abinitha Gourabathina, Zhiyu Wan, J. Thomas Brown, Chao Yan, and Bradley A. Malin — IEEE Transactions on NanoBioscience, 2023
  • PanDa Game: Optimized Privacy-Preserving Publishing of Individual-Level Pandemic Data Based on a Game Theoretic Model
    Abinitha Gourabathina, Zhiyu Wan, J. Thomas Brown, Chao Yan, and Bradley A. Malin — IEEE BIBM 2022
  • What seems to be the problem? Stigmatizing language in patient medical notes
    Abinitha GourabathinaPrinceton University Senior Thesis, 2023