Soham Gadgil
Ph.D. Student in Computer Science & Engineering
Hi! I am a third year CSE PhD student at University of Washington in the AIMS Lab, advised by Dr. Su-In Lee. Prior to starting my Ph.D, I spent a year as a data engineer at Microsoft in the Windows Experience team.
I completed my Masters from Stanford in Computer Science with a depth in AI, where I was a research assistant in the Computational Neuroimage Science Lab (CNSLAB), advised by Dr. Kilian Pohl.
I am also a part-time instructor at Persolv, teaching AI fundamentals to high school students. In my spare time, I like playing tennis, hiking, exploring different cuisines, and watching movies.
Research
The research problems I work on lie at the intersection of Explainable AI and Healthcare:
Transparency in medical AI
- Developed a vision-language foundation model to improve transparency in the medical AI workflow [Nature Medicine 2024]
- Guidelines for building medical AI systems that are transparent and interpretable [Nature Reviews Bioengineering 2025]
Explainable AI
- Developing efficient representation learning methods using sparse autoencoders to interpret the complex behavior of LLMs [Preprint 2025]
- Developed a probabilistic model for feature selection on a per-sample basis in low-resource settings [ICLR 2024]
AI in Healthcare
- Developing unsupervised concept discovery methods for medical AI systems using sparse-autoencoders and LLMs tuned with post-training techniques
- Exploring techniques to fine-tune conditional diffusion models in a low-data regime for steering specific concepts
- Used counterfactual generative models to analyse AI-specific signals enabling classifiers to detect demographic attributes from medical images with high performance [Preprint 2025] [CVPR 2024 DCAMI Workshop]
- Developed a semi-supervised segmentation model for multi-pathology segmentation by combining expert annotations with pseudo-labels [MIDL 2021]
- Formulated a spatio-temporal graph convolution network model for predicting age and sex from MRI scans [MICCAI 2020]
Achievements
- Won the best paper runner-up award at the CVPR 2024 DCAMI Workshop.
- Accepted into the ORS (Opportunity Research Scholars) program for undergraduate research at Georgia Tech.
- Awarded Faculty Honors by Georgia Tech for all semesters in my undergraduate degree.
- Recipient of the Electrical and Computer Engineering (ECE) Senior award for highest academic average.
- Won the best final paper award at Stanford for CS 238.
- Part of the team that first offered the Trustworthy Machine Learning (CS 329T) course at Stanford.
Publications
Preprints
DREAM: A framework for discovering mechanisms underlying AI prediction of protected attributes
Soham Gadgil*, Alex J. DeGrave*, Joseph D. Janizek, Sonnet Xu, Lotanna Nwandu, Fonette Fonjungo, Su-In Lee†, Roxana Daneshjou†
Under Review at Nature Communications
Earlier version presented at CVPR 2024 DCAMI Workshop (Oral, Best Paper Runner-Up)
Peer-Reviewed
[J] Transparency of medical artificial intelligence systems
Chanwoo Kim*, Soham Gadgil*, Su-In Lee
Nature Reviews Bioengineering 2025
[C] Transformer-based Time-Series Biomarker Discovery for COPD Diagnosis
Soham Gadgil, Joshua Galanter, Mohammadreza Negahdar
NeurIPS 2024 TSALM Workshop & ATS 2025
[C] Estimating Conditional Mutual Information for Dynamic Feature Selection
Soham Gadgil*, Ian Covert*, Su-In Lee
ICLR 2024
[C] Data Alignment for Zero-Shot Concept Generation in Dermatology AI
Soham Gadgil, Mahtab Bigverdi
ICLR 2024 DPFM Workshop
[J] Transparent medical image AI via an image–text foundation model grounded in medical literature
Chanwoo Kim, Soham Gadgil, Alex J. DeGrave, Zhuo Ran Cai, Roxana Daneshjou, Su-In Lee
Nature Medicine 2024
[C] CheXseg: Combining Expert Annotations with DNN-generated Saliency Maps for X-ray Segmentation
Soham Gadgil*, Mark Endo*, Emily Wen*, Andrew Y. Ng, Pranav Rajpurkar
MIDL 2021
[C] Spatio-Temporal Graph Convolution for Resting-State fMRI Analysis
Soham Gadgil*, Qingyu Zhao*, Adolf Pfefferbaum*, Edith V. Sullivan, Ehsan Adeli, Kilian M. Pohl
MICCAI 2020
*: equal contribution †: Corresponding Author [J]: journal, [C]: conference
Other Experiences
- Reviewer: ICLR 2025, ICML 2025, NeurIPS 2024, MIDL 2022, ICLR 2024 DPFM Workshop, AISTATS 2025 Good Data Workshop
- President of Georgia Tech IEEE, leading the largest IEEE student branch in the US with over 800 members.
- International Liasion for the Student Alumni Association at Georgia Tech.
- Peer leader in freshmen and senior student dormitories.
Teaching Assistantships
University of Washington
Introduction to AI (CSE 473)
The course covers principal ideas and developments in artificial intelligence: Problem solving and search, game playing, knowledge representation and reasoning, uncertainty, machine learning, natural language processing. I held weekly office hours and assisted in preparing/grading the homework assignments.
Stanford
Computer Organization and Systems (CS 107)
TA for CS 107, one of the largest introductory undergraduate courses at Stanford. I led lab sessions, held office hours, and assisted in grading and designing exams.
Trustworthy Machine Learning (CS 329T)
TA for the first course offering of CS 329T. I co-developed and led the lab sections. I also helped the instructors design lecture slides, homework assignments, and the final project.
Georgia Tech
Linear Algebra (MATH 1554)
As a TA for linear algebra, I led two 50 minute recitation sessions each week.
Computer Architecture (CS 3056)
Guided over 60 students with homeworks and projects in computer architecture. Held weekly office hours, exam review sessions, and collaborated with the instructor for grading and project ideation.