๐Ÿ‘ฉโ€๐Ÿ”ฌ About Me

Hi there! I am currently a Postdoctoral Researcher at NYU Abu Dhabi with Dr. Farah Shamout, working at the intersection of robust vision-language models (VLMs) and agentic systems for clinical decision support. My research focuses on building reliable, interpretable, and adaptive AI systems for high-stakes healthcare settings, with an emphasis on robustness, generalization, and real-world deployment.

I completed my PhD in Computer Vision at the Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI), where I was advised by Dr. Dwarikanath Mahapatra and Dr. Mohammad Yaqub. During my PhD, my work centered on domain generalization, out-of-distribution learning, and improving the robustness of multimodal models in healthcare.

Previously, I interned at the Inception Institute of AI (G42), where I developed adaptive VLM-based solutions for healthcare and built interactive tools for generating customized medical summaries. I have also collaborated with organizations such as Omdena, IBM, and Bose Institute, contributing to projects spanning machine learning, automation, and bioinformatics.

Research Interests: My work is driven by the goal of making AI systems more reliable and meaningful in real-world clinical settings. Broadly, I focus on clinical decision support, medical imaging, and multimodal reasoning as well as:

  • Robustness & Generalization: Handling distribution shifts, OOD scenarios, and test-time adaptation in VLMs
  • Agentic AI Systems: Designing systems that can reason, interact, and assist in clinical workflows
  • Trustworthy AI: Bias, reliability, and interpretability in high-stakes environments

Beyond research, I am interested in the broader questions of ethics, purpose, and responsibility in AI, particularly how they intersect with knowledge traditions and real-world impact.

Feel free to connect if youโ€™d like to collaborate or discuss ideas at the intersection of AI and healthcare!

๐Ÿ”ฅ News

  • 2024.08: ๐ŸŽ‰ Presented the paper accepted at ML4H 2024: "XDT-CXR: Investigating Cross-Disease Transferability in Zero-Shot Binary Classification of Chest X-Rays" held at University of Toronot, Toronot, Canada.
  • 2024.09: ๐ŸŽ‰ Presented the paper accepted at EUVIP 2024: "MRIShift: Disentangled Representation Learning for 3D MRI Lesion Segmentation under Distributional Shifts" at Geneva University Hospita, Geneva, Switzerland.
  • 2024.05: ๐ŸŽ‰ Starting summer internship at Inception Institute of AI (G42), Abu Dhabi
  • 2022.09: Selected as runner-up in SiemensWomenHackAI
  • 2022.03: Served as Jr. Machine Learning Engineer & Instructor at Omdena during Summer of 2022
  • 2021.10: Selected as finalist in MENA Healthcare Hackathon
  • 2021.08: ๐ŸŽ‰ Started PhD in Computer Vision at MBZUAI
  • 2020.01: ๐ŸŽ‰ Joined IBM as Associate Systems Engineer in Bengaluru
  • 2019.12: Presented the paper on pulmonary disease prediction in SIRS Conference
  • 2019.06: Completed my Masters from University of Calcutta, Kolkata, India

๐Ÿ“ Selected Publications

ISBI 2025

Can language-guided unsupervised adaptation improve medical image classification using unpaired images and texts?
Umaima Rahman, Raza Imam, Boulbaba Ben Amor, Mohammad Yaqub, Dwarikanath Mahapatra.

  • Investigating the potential of language guidance in unsupervised domain adaptation for medical image classification using unpaired data.
ML4H 2024

XDT-CXR: Investigating Cross-Disease Transferability in Zero-Shot Binary Classification of Chest X-Rays
Umaima Rahman, Abhishek Basu, Muhammad Uzair Khattak, Aniq Ur Rahman.

  • A comprehensive study on cross-disease knowledge transfer capabilities in zero-shot chest X-ray classification.
EUVIP 2024

MRIShift: Disentangled Representation Learning for 3D MRI Lesion Segmentation under Distributional Shifts
Umaima Rahman, Guangyi Chen, Kun Zhang.

  • Novel approach for handling distributional shifts in 3D MRI lesion segmentation using disentangled representation learning.
SIRS 2019

Obstructive pulmonary disease prediction through heart structure analysis
Umaima Rahman, P. Bhattacharya, Sudipto Saha.

  • Innovative method for predicting obstructive pulmonary disease by analyzing heart structure characteristics.

Demo

๐ŸŽ– Honors and Awards

  • 2022 SiemensWomenHackAI - 1st Runner-up
  • 2022 CISCO Sustainability Challenge, UAE - Among the top 4 finalists
  • 2022 Innovation Challenge for Entrepreneurship - Among the top 8 finalists
  • 2021 MENA Healthcare Hackathon - Among the top 8 finalists

๐Ÿ“– Education

  • ๐ŸŽ“ 2021.08 - Present, PhD in Computer Vision, Mohamed Bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE.
  • ๐ŸŽ“ 2017.08 - 2019.06, Masters (M.Tech.) in Computer Science and Engineering, University of Calcutta, Kolkata, India.
  • ๐ŸŽ“ 2015.08 - 2017.06, Masters (M.Sc.) in Computer Science, University of Calcutta, Kolkata, India.
  • ๐ŸŽ“ 2012.05 - 2015.06, Bachelors in Computer Science, St. Xavierโ€™s College, Kolkata, India.

๐Ÿ’ผ Experience