Mahsa Yarmohammadi

Mahsa Yarmohammadi is an Assistant Research Scientist at the Center for Language and Speech Processing (CLSP), where she leads research in cross-lingual language and speech applications. Her work focuses on deep learning techniques for multilingual data representation, automatic speech recognition (ASR), and speech translation. She develops novel algorithms for ASR decoders and explores interfaces between ASR and machine translation systems.

Education:

• PhD in Computer Science and Engineering, Oregon Health & Science University (2016)
• Master’s in Computer (Software) Engineering, Shahid Beheshti University, Iran (2007)
• Undergraduate degree in Computer (Software) Engineering, Amirkabir University of Technology (Tehran Polytechnic) (2004)

Research Interests: Her expertise spans Natural Language Processing (syntactic/semantic parsing, information extraction), multilingual systems, deep learning, and ASR. She specializes in resource transfer across languages, neural lattice embeddings, and end-to-end system integration for speech and language technologies.

Publication Trends: Recent articles (2019–2025) emphasize multilingual challenges: cross-lingual retrieval, contextual ASR, coreference resolution in dialogue, and zero-shot information extraction. Her work frequently addresses low-resource settings, neural algorithm design, and real-world dataset creation (e.g., earnings call analysis, COVID-19 QA pairs).

Grants & Teams: As a prime researcher in multi-site programs, she collaborates on large-scale multilingual projects. She was mentored by Benjamin Van Durme (Johns Hopkins CS) and Sanjeev Khudanpur (Johns Hopkins ECE) during her CLSP post-doctoral fellowship (2021).