Identifying promising properties

This project will create specialized LLMs that can understand and predict material properties by analyzing vast databases of scientific literature, experimental data and chemical structures to accelerate the discovery-to-application timeline of new materials from decades to months. The system could help researchers identify promising material candidates, predict their properties and suggest optimal processing conditions to support clean energy, electronics and manufacturing applications across industries.

Improving patient outcomes

Researchers are developing a novel deep learning framework that may create more precise prediction models to improve cancer prognosis and outcomes. Unlike current models, this framework captures intratumor heterogeneity by analyzing differences in gene activities across different regions of a tumor to prevent recurrence and therapeutic resistance, delivering more accurate prognostic tools and precise therapies for lung, breast and bladder cancers.

Modeling complex quantum structures 

In this AI-assisted materials design project, researchers will train and use machine learning’s atomic potential in an attempt to model large, multi-layered quantum heterostructures, a task that was deemed cost-prohibitive using traditional methods. Understanding these structures could advance the design of functional materials used in the energy sector such as rare-earth-free permanent magnets and high-performance photovoltaic materials.

Disordered alloys and oxides

Disordered alloys and oxides are materials in which different atoms can occupy many possible arrangements, creating a design space too vast to explore with traditional simulations. This project will use AI models to learn how composition and local atomic structure affect material stability and properties, then guide the search toward promising atomic arrangements. Empire AI’s high-performance GPU computing allows teams to train and test these models at scale, potentially accelerating the discovery of materials for batteries, clean-energy technologies and advanced manufacturing.

Interactive model for children

UB researchers will utilize Empire AI to build an agentic, multimodal large language model to perform real-time, interactive speech-language therapies for children with speech and language challenges. This model will be used by the National AI Institute for Exceptional Education and is designed to plan, deliver, and adapt interventions in real-time based on the child’s responses and scene dynamics.

Data-driven financial forecasting

The goal of this Financial Topic Attention model is to deliver data-driven, actionable intelligence essential for financial forecasting, asset pricing and risk management. By integrating large language models with clustering and refinement algorithms, this project aims to solve contextual limitations in current systems to provide a structure that produces more coherent, interpretable and predictive insights.

AI-driven 3D modeling 

This project combines advanced AI with physics-based modeling to transform standard brain blood flow images into detailed 3D simulations. By integrating medical imaging and computational physics, the research seeks to capture more nuanced patterns in how blood moves through neurovascular systems. This approach has the potential to generate faster, more individualized insights that could one day support physicians in understanding complex brain vessel conditions.

AI-enhanced medical imaging

A multidisciplinary team is creating an AI tool that may help doctors better diagnose and monitor diseases. It will generate biomedical images by combining pathology reports with image-generating AI models that are trained on medical scans. The tool could lead to more sophisticated models that integrate genetic data with medical imaging.

Optimizing plant growth 

Researchers are training AI on quantum fingerprints (patterns in how quantum light arrives at a detector) to pick out faint optical signals inside living plants that are normally drowned out by background light from chlorophyll, the pigment that gives plants their green color. This lets scientists and growers precisely measure how crops respond to light conditions, nutrients and environmental cues to tailor interventions, which could improve algae-based carbon capture, increase agricultural yields, contribute to resilient food systems and boost sustainable bio-manufacturing.

Faster, more accurate detection 

This project aims to develop a powerful AI foundation model designed specifically for analyzing histopathology images—microscopic views of tissue samples used to diagnose diseases such as cancer, autoimmune disorders and infections. By training on large, diverse datasets and leveraging high performance GPU clusters, the model may enable applications ranging from disease classification and automated reporting to advanced image search and retrieval with the potential to advance early detection and improve diagnostic accuracy. 

Generative AI evaluation

While generative AI and large language models have shown promise in clinical research and decision support, their performance in these settings has not been widely examined. This project will evaluate the complex reasoning skills of generative AI and LLMs by studying deep-learning models of neurological diseases like multiple sclerosis, Alzheimer’s disease and opioid use disorder, generative AI for virtual populations and LLMs for clinical trial protocols and pharmacy counselling.

Identifying children’s unique needs

This project is developing an AI-powered screening tool that can interpret multiple types of information—such as speech, text and visual cues—to identify children’s unique learning and developmental needs. By training an LLM on a rich, multimodal dataset, the system is designed to recognize patterns in communication and behavior that traditional assessments may not easily detect, allowing educators and specialists to tailor interventions and strategies for each child.

Answering healthcare questions

Building on previous research to optimize LLM’s ability to answer medical questions, this project will fine-tune parameters to facilitate semantic reasoning by training the model on millions of facts pertinent to health and healthcare and integrating added layers of reasoning. Semantic AI could provide more accurate and secure access to LLMs to support medical practice, education and research.

Clearer, faster diagnosis

A SPECT scan uses a radioactive tracer and a specialized camera to create 3D images that reveal how organs and tissues are functioning. Current scanners face an inherent trade-off: sharper images come at the cost of detecting fewer tracer signals, while capturing more signals reduces image sharpness. Empire AI is helping accelerate the development of a new SPECT scanner designed to potentially overcome this trade-off, with the goal of

Securing software development 

In an effort to reduce persistent security vulnerabilities in AI-assisted software development based on natural language, this project harnesses human logic to bolster the integrity of cybersecurity tools used across industries. By applying Structured Secure Coding Practices (SCPs), this proposed model aims to significantly reduce security flaws found in functional code generated by large language models and autonomous coding agents.