Research Interests

Natural Language Processing · Computational Linguistics · Multimodal AI

Natural Language Processing

My primary research interest lies in understanding and advancing how machines process human language. I'm particularly fascinated by the intersection of linguistics and deep learning.

LLM Interpretability & Mechanistic Interpretability

Understanding how large language models internally represent linguistic structures, with a focus on probing attention mechanisms and neuron activation patterns for syntactic and semantic features.

Low-Resource Language Modeling

Developing efficient NLP techniques for languages with limited training data, with emphasis on transfer learning and cross-lingual representation learning for underrepresented languages.

Spoken Language Understanding

Exploring end-to-end models for conversational AI that can simultaneously process acoustic signals and linguistic content for more natural human-computer interaction.

Computational Linguistics

Syntax-Semantics Interface

Investigating how formal grammars and semantic representations can be integrated with neural models to improve compositional generalization in language understanding tasks.

Language Change & Evolution

Using computational methods to track semantic drift and linguistic evolution across time, with applications in digital humanities and historical linguistics.

Multimodal AI

Exploring the intersection of language and vision through:

  • Vision-Language Pretraining — Developing models that learn joint representations from images and text for tasks like visual question answering and image captioning.
  • Audio-Visual Speech Recognition — Leveraging visual cues from lip movements to improve speech recognition in noisy environments.

Cross-modal
representation learning

Current Research Directions

Probing syntactic knowledge in LLMs
Cross-lingual transfer for low-resource languages
Multimodal reasoning for educational tools
Computational analysis of linguistic change