Research
Research interests
Areas of interest
Large language models
How LLMs behave on low-resource languages, and how to use them reliably in real applications.
Multilingual embeddings
Representing text from different languages in a shared vector space so meaning can be compared across languages.
Nepali language processing
Tools, datasets and models for Nepali, a language that is still under-served by mainstream NLP.
Natural language processing
Core NLP problems such as tokenisation, classification and retrieval, with a practical engineering focus.
Semantic search
Search that matches meaning rather than exact keywords, built on embeddings and vector retrieval.
Multilingual information retrieval
Finding relevant documents when the query and the documents may be in different languages, such as Nepali and English.
Publications
No publications yet. Papers, preprints and datasets will be listed here with their abstracts, DOIs, code and citation information as they are released.