About

I am a postdoctoral researcher at Microsoft in the Office of Applied Research working with Jaime Teevan, Brent Hecht, and Bahareh Sarrafzadeh. In May 2024, I graduated with a Ph.D. in computer science from the University of Massachusetts Amherst, co-advised by Andrew McCallum and Hamed Zamani.

My research sits at the intersection of Natural Language Processing, Information Retrieval, and Human-Computer Interaction. I develop human-in-the-loop AI models that collaborate with knowledge workers in their professional workflows where they find, consume, and produce information. I develop models which enable rich interaction with users as well as models that learn from users' historical interactions to personalize responses to them.

My work is often user-centered and tackles realistic, yet under-supported challenges in knowledge work. In the past, I have worked on:

• Interactive and personalized ranking models that make reviewer-paper assignments in peer review.
• Interactive exploration models and datasets to help scientists in brainstorming with the literature.
• LLMs personalized with calibrated retrieval models for writing assistance.
• Low resource information extraction models and datasets to help materials scientists to analyze materials science papers to automate parts of materials synthesis.

I am drawn to methods which enable models to communicate and interact with humans:

• Explainable models which communicate model internals with users e.g. LACE, CtrlCE
• Calibrated models which communicate uncertainty in predictions or intermediate states e.g. Pearl, CtrlCE
• Optimization methods to incorporate inference time feedback into predictions e.g. LACE, EdTM
• Data-centric methods like weak supervision and synthetic data that enable training from imperfect human annotations e.g. Aspire, MINT

Finally, datasets enable model development and user-centered analysis point to meaningful problems which may be tackled in technical work. So I have developed important datasets for under-explored problems: LAMP, CSFCube, MSPT. And conducted human-centered analysis to understand user behaviors and needs: Lit-review behaviors, User-LLM collaboration behaviors.

You can see my CV here: link

Or visit me here: Twitter | Linkedin | Google Scholar | Semantic Scholar

Or send me an email if you want to speak to me.