Seeking research internships, 2027

Mahsa Khoshnoodi

PhD Student, Computer Science · Georgetown University

I work on multimodal AI, studying how models see and reason about the world, and building the tools that reveal when that understanding is real.

GUCV Lab · Advised by Prof. Sarah Adel Bargal, mentored by Dr. Michael Saxon.

I study where and why vision-language models fail on fine-grained visual understanding. Rather than treating hallucination as an output artifact, I build diagnostic frameworks that trace it back to the exact point in a model's reasoning where perception breaks down. My longer view: capable multimodal systems will need a structured world model that connects seeing, understanding, and acting.

Research directions

Perception to reasoning

I investigate how VLMs integrate visual and linguistic information to reach decisions. Even when models land on correct conclusions, their internal reasoning paths are often flawed or biased. I build interpretability tools that act as a microscope for AI, tracing information flow and exposing where perception fails to become genuine reasoning.

Diagnostic frameworks for VLMs

I develop evaluation frameworks that assess not just whether a model is correct, but whether its reasoning process is valid. Hallucination and bias appear heterogeneously across layers and architectures, so effective diagnosis means reading internal dynamics, not just observing outputs.

From seeing to acting

My long-term work targets systems that perceive, reason, and act reliably in the world. Drawing on structured world modeling and vision-language-action architectures, I aim to build multimodal systems that stay aligned with human values across the full loop from visual input to real-world decision.

Selected publications
Paper 2026
Rupayan Mallick, Mahsa Khoshnoodi, Sarah Adel Bargal
CVPR-VisCon 2026
Mahsa Khoshnoodi, Sarah Adel Bargal
Third Workshop on Visual Concepts (VisCon), CVPR 2026
SemEval 2026
Mahsa Khoshnoodi*, Rojin Ziaei*, Nazli Goharian* Equal contribution
SemEval 2026
KDD 2025
Hierarchical Prompting Taxonomy: A Universal Evaluation Framework for Large Language Models
Devichand Budagam, Sankalp KJ, Mahsa Khoshnoodi, Ashutosh Kumar, Vinija Jain, Aman Chadha
KDD 2025
NeurIPS 2024 Spotlight, top 5%
Who Evaluates the Evaluations? Objectively Scoring Text-to-Image Prompt Coherence Metrics with T2IScoreScore (TS2)
Mahsa Khoshnoodi*, Fatima Jahara*, Michael Saxon*, Yujie Lu, Aditya Sharma, William Yang Wang* Equal contribution
NeurIPS 2024
Paper 2024
A Comprehensive Survey of Accelerated Generation Techniques in Large Language Models
Mahsa Khoshnoodi, Vinija Jain, Mingye Gao, Malavika Srikanth, Aman Chadha
Preprint, 2024
Under review
Reimagining Neurosymbolic AI through the Lens of Cognitive Science: A Survey
Devichand Budagam, Mahsa Khoshnoodi, Jibesh Patra, Ravid Shwartz-Ziv, Amit Sheth, Vinija Jain, Aman Chadha
Under review, ACM Computing Surveys
News
May 2026
Paper accepted at CVPR 2026 VisCon Workshop: "Do VLMs Reason About Faces?"
Apr 2026
Paper accepted at SemEval 2026: Emotion-Aware Multi-Task Learning for Conspiracy Detection.
Aug 2025
Started PhD at Georgetown University, joining the GUCV Lab under Dr. Sarah Adel Bargal.
Feb 2025
KDD 2025: Hierarchical Prompting Taxonomy paper published.
Dec 2024
NeurIPS Spotlight: T2IScoreScore recognized as a Spotlight paper (top 5%).