I’m an MS Computer Science student at the George Washington University, and a visiting student at New York University working with Prof. Duygu Ataman. I have co-founded aLLMA Lab to promote NLP research for Azerbaijani language. My current research interests include multilingual language modeling and adversarial attacks on LLMs. So far, I have had a chance to work on the following problems:

  • Neural spelling correction for agglutinative languages.
  • Building web-scale text corpus and monolingual LLMs for Azerbaijani.
  • Developing contex-sensitive retrieval-augmented generation.
  • Benchmarking SOTA LLMs’ NLU capabilities on Turkic languages.
  • Prompt compression via adversarial attack algorithms.

Before joining NYU, I was a lead machine learning engineer at PRODATA LLC, where I lead multiple industry projects on customer chatbots. Before that, I was a machine learning engineer at Azerbaijan AI Lab (within Azercosmos), where I developed spelling correction software for internal use of certain government agencies. I completed my bachelor’s degree on Biomedical Engineering at Azerbaijan State Oil and Industry University.

We have released TUMLU: A Unified and Native Language Understanding Benchmark for Turkic Languages. The preprint is available on arXiv. Code and data are hosted on GitHub.

News

Feb 2025 Our paper, TUMLU, is now available on arXiv.
Dec 2024 I was a reviewer for MRL Workshop at EMNLP 2024 in Miami, Florida.
Aug 2024 Joined NYU as a visiting student to work with Prof. Duygu Ataman.
Aug 2024 Our work on Azerbaijani foundation models received Best Paper Award (Honorable Mention) at SIGTURK Workshop/ACL 2024.
Jul 2024 Our work on contextual document retrieval was accepted to IEEE AICT 2024.
Jun 2024 Our work on Azerbaijani foundation models was accepted to SIGTURK Workshop at ACL 2024.

Older news.