Applied Scientist
Email: jam <at> lissus <dot> com
Download CV
About me
I'm an Applied Scientist at Adobe on the Adobe Experience Platform GenAI team.
Previously, I did my Ph.D. in computer science at Stony Brook University working on
belief and cognitive state modeling in text and speech. I was advised by Owen Rambow.
My work broadly spans computational linguistics and cognitive science. I have worked on bridging
theory-of-mind concepts with modern NLP. I am also interested in large language model's ability
to detect hedges, sarcasm, or other subtle markers of speaker intent. Some of this work is a joint and ongoing collaboration with Susan Brennan from the Stony Brook Psychology Department.
Updates
May 2025
Successfully defended my Ph.D.! I will begin at Adobe on 5/27 and will be moving to San Francisco.
January 2025
My Adobe internship work on Enhanced Clarification for Interactive Responses (ECLAIR)
to appear AAAI 2025 (Demo) and IAAI 2025. Another publication on Synthetic Audio Data
accepted to Findings of NAACL 2025.
May - Nov. 2024
Interned at
Adobe
as an ML Engineer, building an interactive disambiguation pipeline for the AEP AI assistant.
I had an excellent experience, and was mentored by
Zifan Liu and manager
Yunyao Li. Two papers accepted and a patent submitted!
Sep. - Dec 2023
Interned at the Genius Institute at
Gap International
as an NLP Engineer. I helped leverage LLMs to detect patterns of genius and built an
unsupervised clustering pipeline for genius detection. I was mentored by my amazing father
Alex Murzaku (also fulfilling my childhood
dream of one day working together with my dad).
Publications
-
OmniVox: Zero-Shot Emotion Recognition with Omni-LLMs
-
Zero-Shot Belief: A Hard Problem for LLMs
-
Projecting Knowledge and Common Ground from Characters’ Utterances in Narratives: A Psycholinguistic Baseline for LLMs
ToM4AI Workshop (AAAI 2025)
A. Soubki, A. Paige,
J. Murzaku,
O. Rambow, S. Brennan
Psychology Dept. Collaboration
-
ECLAIR: Enhanced Clarification for Interactive Responses in an Enterprise AI Assistant
-
ECLAIR: Enhanced Clarification for Interactive Responses
-
Synthetic Audio Helps for Cognitive State Tasks
-
Training LLMs to Recognize Hedges in Spontaneous Narratives
-
Views Are My Own, but Also Yours: Benchmarking Theory of Mind Using Common Ground
-
Multimodal Belief Prediction
-
BeLeaf: Belief Prediction as Tree Generation
-
Towards Generative Event Factuality Prediction
-
Re-Examining Factbank: Predicting the Author's Presentation of Factuality
* denotes equal contribution