Two CS&E PhD Students Earn Interdisciplinary Doctoral Fellowships

Department of Computer Science & Engineering PhD students Shirley Anugrah Hayati and Azal Ahmad Khan have earned an Interdisciplinary Doctoral Fellowship for the 2026-27 academic year. Each recipient will receive a $25,000 stipend, academic year tuition at the general graduate rate for up to 14 credits per semester, and subsidized health insurance through the Graduate Assistant Health Plan for up to one calendar year.

Awarded by the University of Minnesota Graduate School, the Interdisciplinary Doctoral Fellowship provides a unique opportunity for outstanding mid-career Ph.D. students who are engaged in interdisciplinary research to study with faculty at one of the University’s interdisciplinary research centers or institutes during the fellowship year. Only 18.5% of nominees are selected for the award. 

Learn more about Hayati and Khan’s research below!

Shirley Anugrah Hayati

Advisor: Dongyeop Kang
Interdisciplinary Research Center/Institute: Data Science and AI Hub (Josef Woldense)

Social diversity refers to differences in how people think, behave, and interact as shaped by their social contexts. In humans, social diversity manifests across multiple layers, including perspectives, values, goals, preferences, and behaviors. While recent advancements in large language models (LLMs) have demonstrated strong capabilities in reasoning and natural language generation, these models often fail to reflect the full diversity of human perspectives and behavior. Capturing social diversity is critical for making fair decisions, conducting trustworthy social simulations, and supporting effective human-LLM collaboration. Integrating insights from computer science, linguistics, and social science, Hayati’s interdisciplinary research develops novel methods and frameworks that enable LLMs to represent diverse viewpoints, personal traits, contextual factors, and behavioral patterns for more pluralistic AI systems. By modeling diversity in how people think and interact, these systems can better align with human needs and be applied more safely in complex real-world social applications.

Azal Ahmad Khan

Advisor: Ali Anwar
Interdisciplinary Research Center/Institute: Minnesota Institute for Astrophysics (Michael Coughlin)

Khan works on improving the efficiency of machine learning (ML) systems, with an emphasis on large language models and agentic, multi-step reasoning workflows. The central problem he focuses on is that most ML systems become expensive or slow when run at scale or under real-world constraints. His research aims to close this gap between capability and practicality. On the inference side, he designs methods that reduce the compute required for reasoning and agentic tasks by cutting unnecessary generation and exploiting structure, such as query similarity, while preserving accuracy on standard benchmarks. On the training side, he builds execution-aware fine-tuning systems, including adaptive controllers that tune configuration choices based on runtime signals, enabling training effectiveness and efficiency. Next, he plans to scale these ideas into compound AI systems where many specialized agents collaborate, with a focus on interdisciplinary applications that require reliability, reproducibility, and measurable efficiency.

Congratulations to Shirley Anugrah Hayati and Azal Ahmad Khan on this accomplishment!