Harsh Nishant Lalai
Incoming PhD Student at
Johns Hopkins
Hey there! I'm an incoming PhD student in Computer Science at
Johns Hopkins University,
co-advised by
Kristina Gligorić
and
Benjamin Van Durme.
Previously, I completed my undergraduate degree at
BITS Pilani, Goa Campus.
Johns Hopkins University,
co-advised by
Kristina Gligorić
and
Benjamin Van Durme.
Previously, I completed my undergraduate degree at
BITS Pilani, Goa Campus.
I'm interested in trustworthy AI, focusing on language models under
uncertainty and multi-agent social interactions. My goal is to develop
rigorous evaluations and robust methodologies that enable increasingly
capable AI systems to be deployed safely and reliably in real-world settings.
I've collaborated with researchers at
Stanford,
Harvard,
Georgia Institute of Technology,
Pennsylvania State,
and
Emory University.
Stanford,
Harvard,
Georgia Institute of Technology,
Pennsylvania State,
and
Emory University.
Featured Research Publications
A Comprehensive Taxonomy and Challenges in Text Watermarking
for Large Language Models
Findings of NAACL 2025
How Geographic Origin Influences LLMs' Entity Deduction
Capabilities
COLM 2025
Evaluating Vision-Language Models on Misleading
Data Visualizations
arXiv 2026
Updates!
Excited to start my PhD at
Johns Hopkins University
in Fall 2026, co-advised by
Kristina Gligoric
and
Benjamin Van Durme!
When Visuals Aren't the Problem
is now available on arXiv.
The World According to LLMs was accepted to COLM 2025.
Started my undergraduate thesis at
Georgia Institute of Technology
under the advisement of
Sashank Varma
and
Raj Sanjay Shah.
From Intentions to Techniques was accepted to Findings of NAACL 2025.