Computer science grad earns IBM fellowship for AI research
Computer science grad earns IBM fellowship for AI research
Publish Date: 2026-04-30 18:09:00
Source Domain: news.asu.edu
Modern medicine runs on data organized into tables and charts that help doctors make fast, informed decisions. Survival rates and prescription side effects are often distilled into formats that can be scanned in seconds. But that clarity doesn’t come easily. The data must be extracted and organized, often through a time-consuming process that can leave critical information buried deep in research reports.
Naman Ahuja wants to change that.
This May, Ahuja will graduate from the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at Arizona State University, earning a master’s degree in computer science.
Over the past two years, his research has focused on how to get artificial intelligence, or AI, systems to convert long, unstructured text into accurate, usable tables. This spring, that work earned him an IBM Infrastructure Master’s Fellowship Award, a prestigious honor that recognizes research with strong real-world and industry impact.
Looks right, isn’t right
The problem exposes a key limitation of modern AI. Large language models can read and summarize documents with ease. But when asked to extract precise information and organize it into something structured — like a table a doctor or analyst could rely on — they often struggle. Important details can be missed, information can become inconsistent, and models may generate claims that aren’t supported by the original text.
The result can look polished, but it isn’t always reliable. Ahuja’s research focuses on closing that gap.
“In the real world, a lot of data exists in complex and semi-structured formats, like PDF documents or Wikipedia pages,” he says. “These documents have some structure, but they’re still complex and contain a lot of information.”
His solution, developed through his master’s thesis, rethinks how AI should approach the problem.
Instead of asking a model to generate a table in one pass, Ahuja breaks the task…
