2025: Multiple NLP papers in preparation or under review with collaborators including David M. Rothschild and Kathleen McKeown; one accepted to EMNLP Main Proceedings.
June 2025: Lead inventor on U.S. Patent US 12,321,701 (application 17/981293).
2023: Paper accepted at WWW’23 (acceptance rate 19.8%).
2022: Ranked #1 in NIST TRECVid DVU task on Multimodal Language Understanding.
2021: Ranked #1 in ACM Grand Challenge on Multimodal Semantic Understanding in long-form datasets.
2020: Published three peer-reviewed papers through external collaborations.
2019: Honored as Class Marshal of Columbia University Class of ’19 for academic excellence.
Published works include 'iBERT: Interpretable Style Embeddings via Sense Decomposition' (with Kathleen McKeown) and 'MultiSeg' (with Smaranda Muresan et al., LREC 2020).
Developed 'Low-Resource Interactive Visualization' demo (LREC 2020).
Research Experience
Senior Applied Scientist at Microsoft, leading research in Generative AI, Conversational AI, and Multimodal systems.
Owns the Smart-Reply project and contributes to AI components in the Copilot project.
Served as Research Mentor for an MSR NLP PhD Summer Intern in 2022.
Conducted research on multimodal semantic inference in long-form datasets; delivered invited talks at NIST and Columbia.
Previously contributed to Uber AI, recognized in their official blog.
Background
Senior Applied Scientist at Microsoft working in the Generative AI space.
Leads the Smart-Reply Microsoft Research project (est. 2020), saving ~294,000 hours/year and ~$3.6M annually.
Research focuses on long-tailed domain-specific Conversational AI LLMs and Multimodal AI in the Copilot project.
Collaborates with the Computational Social Science group at Microsoft Research NYC.
Shares insights from Big-Tech research experience through an AI Blog.
Miscellany
Advisory board member of an NGO since 2021, focusing on AI-based education for underprivileged slum children.