AI Researcher
PhD Candidate · Indiana University Bloomington
I study how language models represent, transfer, and reason about linguistic information across speech, text, and multilingual settings. My work spans mechanistic interpretability, natural language inference, and knowledge-grounded AI.
I am a dual-major PhD candidate in Computational Linguistics and Middle Eastern Languages & Cultures at Indiana University Bloomington, with a minor in Computer Science. My research centers on mechanistic interpretability, multilingual natural language inference, and evidence-grounded AI systems.
I investigate what language models encode internally, how reliably they transfer across languages, and how structured knowledge can ground their outputs. I also build public datasets and tools, including Rasid and the cloud-based research editor RogueTeX.
Connecting linguistic theory with empirical machine learning to build AI systems that are interpretable, reliable, and grounded in evidence.
Probing internal model representations to understand how speech and language transformers encode linguistic features, nativeness, and structure.
Studying entailment, contradiction, and pragmatic inference across languages, with a focus on evaluation reliability and label drift introduced by machine translation.
Building ontology-based knowledge graphs and retrieval systems that turn complex scientific literature into structured, queryable, and evidence-grounded resources.
Developing datasets, corpora, and models for languages and domains that remain underrepresented in modern AI, including work on variation, ellipsis, entities, and discourse.