Empire AI powers bold new research at UB, from surgery to climate science

With Empire AI gearing up for its second phase, UB researchers are already harnessing its supercomputing system to address pressing societal challenges.

UB scholars are leading a new wave of artificial intelligence-driven projects that aim to provide doctors with advanced tools for brain and gastrointestinal surgeries, help farmers create more sustainable food production systems and supply policymakers with more nuanced data on financial markets.

“At the University at Buffalo, we are dedicated to harnessing artificial intelligence for the public good, and Empire AI is accelerating this mission,” says Venu Govindaraju, senior vice president for research, innovation and economic development, and Empire AI board member. “By providing the computational power once limited to private tech companies, Empire AI empowers researchers at UB and across New York State to continuously push the boundaries of discovery and drive innovations that improve lives and drive economic growth.”

The projects and others described below follow a batch of Empire AI-driven research led by UB faculty members earlier this year. Each project supports the mission of Empire AI, which is harnessing AI for the betterment of society and driving innovation in New York State.

Launched in 2024 by Gov. Kathy Hochul, Empire AI is a research consortium backed by more than $500 million in public and private funding. It consists of 10 universities, including UB and other SUNY institutions, as well as several private research institutions across the state. A SUNY flagship with decades of leadership in AI and data science, UB is also home to Empire AI’s supercomputing center.

The supercomputing center is about to receive a boost with the upcoming launch of Empire AI Beta, its second phase that will be 11 times more powerful than current capacity.

Here’s a closer look at the second round of UB projects made possible by Empire AI.

This project centers on creating an AI system that functions as a digital expert in agriculture and climate science, says principal investigator David Doermann, SUNY Empire Innovation Professor and chair of the Department of Computer Science and Engineering.

Doermann will employ Empire AI’s supercomputers to train a large language model to recognize nuanced agriculture and climate concepts, as well as link visual data with complex climate and farming questions. Similar multimodal tools exist, he says, but many miss critical details in evolving fields like agriculture and climate science.

“Ultimately, we aim to develop tools to support farmers, researchers and policymakers in their work creating sustainable food production systems and climate resilience,” says Doermann.

The project, led by Ciprian Ionita, associate professor of biomedical engineering, integrates advanced medical imaging with physics-informed AI to model and analyze blood flow dynamics in the brain.

Using Empire AI’s computational resources, Ionita’s team will train and validate neural networks that reconstruct full 3D hemodynamic simulations consistent with real angiographic measurements and governed by the fundamental equations of fluid motion and contrast transport.

This approach enables generation of quantitative circulation maps that extend beyond what can be measured from imaging alone, providing clinicians with new tools to assess subtle vascular abnormalities, guide endovascular treatments and improve mechanistic understanding of cerebral blood-flow behavior.

Cristian Tiu, associate professor of finance, is principal investigator of this project, which will leverage Empire AI’s computing power to explore how massive amounts of information move through financial markets.

The first part aims to leverage generative AI to predict trade patterns, similar to how large language models predict the next word. “Trades and quotes during a day are analogous to a text, and the next trade in a sequence is analogous to the next word in a sentence,” Tiu says.

The second part employs AI to mine insights from historical financial news and reports, while avoiding look‑ahead bias inherent in many of today’s AI models.

Ultimately, the work aims to improve predictive models, uncover hidden patterns in trading behavior, and help investors and researchers better process the immense flow of market information.

This project aims to develop an AI “co-pilot” that combines visible and near-infrared imaging to support surgeons during common gastrointestinal surgeries, says principal investigator Peter Kim, professor and vice chair for research and innovation in the Department of Surgery.

Working with UB computer scientists Jinjun Xiong and Junsong Yuan, Kim will employ Empire AI to develop multimodal large language models that combine visible and near-infrared imaging to support surgeons during common gastrointestinal surgeries involving the liver and colon.

He will also develop AI tools that improve operating room efficiency and safety, including identifying sterile surgical instruments in real time; analyze electronic medical records; and assist with audio communication, enabling faster and more accurate emergency decisions. 

This project, part of the National AI Institute for Exceptional Education at UB, will enhance the institute’s AI Screener, which works to identify children in need of speech and language support.

The effort is led by Jinjun Xiong, SUNY Empire Innovation Professor of computer science and engineering, and director of the UB Institute for Artificial Intelligence and Data Science.

He will utilize Empire AI to train a new foundational model that will refine the screener, which interprets multiple types of information — speech, written word, visual cues and more — that traditional screening tools assessments may not easily process and detect.

The goal, Xiong says, is to support educators and specialists in tailoring interventions and strategies for each child, with the potential for personalized learning.