VISION Use Cases

VISION in Action: Real-World Impact

VISION is already powering cutting-edge research across a wide range of disciplines. The projects featured here are actively using the supercomputer's advanced capabilities to tackle complex challenges that would be impossible to address with conventional computing resources.

Accelerating Drug Discovery Through Large-Scale Screening

Dr. Reid T. Powell, Texas A&M University

Dr. Reid T. Powell leads a drug discovery research group dedicated to identifying novel compounds that could advance the development of cancer treatments. His team recently completed a large-scale virtual screening of approximately 10.4 million compounds using advanced computational models. The run, which consumed over 26,000 GPU-hours, was completed in about one week — work that would have taken years in their previous environment.

This effort identified more than 22,000 potential candidate molecules across multiple binding regions, significantly accelerating the path from target identification to experimental discovery. Dr. Powell reported that the system performed exceptionally well, achieving nearly complete dataset coverage and demonstrating computational capacity comparable to that of major pharmaceutical environments.

Why Vision Was Built: The Case for Investment

When making the case for investing in VISION, researchers across Texas A&M identified critical challenges that demand massive computational power. These examples illustrate the scope and diversity of problems that VISION is designed to address—from climate modeling to healthcare AI to disaster response. They represent the possibilities that inspired VISION's creation.

Addressing Climate Uncertainties with AI and HPC

Dr. Ping Chang, Texas A&M University

Dr. Chang's team is working to improve how scientists predict extreme weather events like hurricanes, heavy rainfall, and sea-level rise by combining artificial intelligence with highly detailed climate simulations. Current climate models often lack the precision needed to accurately forecast these dangerous events at the local level, where communities need actionable information. By running simulations at much higher resolution with VISION, researchers could better capture the complex interactions between the ocean and atmosphere that drive extreme weather. This work directly responds to a 2023 White House science advisory report calling for more accurate climate projections to help communities prepare for a changing climate.

VISION's powerful computing capabilities would allow Dr. Chang's team to process the enormous amounts of data required to train AI models and run detailed simulations that would otherwise take years to complete. The results would help policymakers, emergency managers and communities make better decisions about infrastructure, evacuation planning and long-term climate adaptation.

Decision-Making in Healthcare

Dr. Yu Zhang, Texas A&M University

Dr. Zhang is developing AI systems that can analyze medical images and clinical notes together to help doctors make better decisions about patient diagnosis and treatment. These "vision-language" models learn to understand the relationship between what appears in medical scans and what physicians write in patient records, potentially catching patterns that humans might miss. Such technology could help identify diseases earlier, recommend more personalized treatment plans, and summarize complex medical findings for both doctors and patients. Training these sophisticated AI models requires processing millions of medical images and documents, which demands immense computing power.

VISION would enable Dr. Zhang's team to build and refine these models at a scale that would be impossible with standard computing resources. The potential impact includes faster, more accurate diagnoses and reduced healthcare costs, particularly benefiting patients in underserved areas who may have limited access to specialists.

Trustworthy Large Language Models

Dr. Kuan-Hao Huang and Dr. Tomer Galanti, Texas A&M University

Dr. Huang is researching how to make AI chatbots and writing assistants more reliable by preventing them from producing harmful, biased or factually incorrect responses. Large language models like ChatGPT learn from vast amounts of internet text, which can inadvertently teach them problematic patterns or misinformation. Understanding exactly how training data influences AI behavior requires building and comparing many different versions of these models with carefully varied data inputs. This type of systematic experimentation is only possible with access to large numbers of powerful processors working simultaneously.

VISION would provide the computational scale needed to train dozens of experimental AI models in parallel, dramatically accelerating this critical safety research. The findings will help ensure that AI tools used in education, business, healthcare and everyday life are more trustworthy and aligned with human values.

Leveraging Technology for Disaster Management

Dr. Zhe Zhang, Texas A&M University (NSF CAREER Awardee 2023)

Dr. Zhang is building intelligent mapping and decision-support tools to help communities better prepare for and respond to flooding disasters. By combining AI with geographic data, his team creates detailed flood-risk maps and vulnerability assessments that help emergency managers identify which neighborhoods and populations are at the greatest risk. These tools are designed to be transparent and explainable, helping citizens understand their personal flood risk rather than just presenting abstract data. The system processes satellite imagery, sensor data, and geographic information in real time, requiring substantial computing power to deliver timely results during emergencies.

VISION would enable the rapid processing needed to update flood predictions as storms develop and to run complex simulations that account for many variables simultaneously. Through partnerships with NOAA and other federal agencies, these tools will be made available to emergency managers nationwide, with particular focus on helping vulnerable and underserved communities.

AI-Powered Health Coaching for Adolescents

Dr. Xin Li, Texas A&M University

Dr. Li is creating a digital health system featuring an AI-powered virtual coach that encourages teenagers to exercise more and eat healthier through personalized conversations and a lifelike animated avatar. The system analyzes each user's activity patterns and dietary habits to provide tailored advice and motivation, similar to having a personal health coach available anytime on a smartphone. Creating realistic virtual humans capable of natural conversations requires training AI models on massive amounts of data on human movement, speech and behavior. Companies developing similar technology typically use thousands of specialized processors running for weeks or months to train a single model.

VISION would dramatically reduce this training time, allowing Dr. Li's team to experiment with different approaches and refine the virtual coach more quickly. This research could help address the adolescent obesity epidemic by making personalized health guidance accessible and engaging for young people who might not otherwise have access to nutrition counseling or fitness coaching.