April 2026
Friends,
Our centerpoint VALIANT PITCH session at Discover AI 2026 delivered exactly what we hoped for and more: a fast-moving, high-energy showcase of ideas spanning disciplines, career stages, and levels of ambition. As we opened the session, the tone was clear. This was not about a single vision of AI. It was about building a community where ideas collide and scale together.
Across 13 rapid-fire pitches, teams were given just three minutes to articulate problems, propose solutions, and make the case for investment. The result was a remarkable cross-section of innovation. We saw everything from neuro-fuzzy control systems for lunar construction to AI-powered clinical registries that could transform emergency surgery workflows. Teams pushed into new frontiers of imaging with low-field MRI, reimagined surgical guidance through simulation-driven AI, and tackled real-world infrastructure challenges such as grid resilience through electric vehicles.
What stood out was not just technical creativity, but translational intent. Many teams framed their work around deployment: reducing healthcare costs, improving educational outcomes, enabling accessibility, or reshaping everyday experiences like travel, retail pickup, and recycling. The emphasis on actionable impact echoed throughout the session.
The awards reflected both technical depth and real-world potential. Four teams were selected as equal winners (profiled later in this letter): SPINS, a visual dictionary for neuroscience that brings clarity to complex brain atlases; Resilio, a platform turning electric vehicles into on-demand grid assets during outages; Edge MRI, advancing low-field imaging with AI-enhanced reconstruction; and LANTERN, an AI-powered reading assessment that translates student thinking into actionable insights for educators. In addition, SPINS captured the Audience Choice Award, highlighting its immediate resonance with the broader community.
Equally important was the format itself. Judges brought industry perspective and rapid questioning, while the audience played an active role through real-time feedback and voting. This created a dynamic loop where ideas were not only presented but stress-tested and refined in the moment.
Perhaps most compelling was the diversity of voices. High school students, undergraduates, graduate trainees, and faculty stood side by side, each contributing distinct perspectives. This is the VALIANT model in action: not a vertical silo, but a horizontal network of thinkers building together.
As the session closed and conversations spilled into AI demos in atrium, one thing was clear. When we #DiscoverAI together, it is not just a competition. It is a catalyst.
Bennett
DiscoverAI in Action
DiscoverAI Demos filled the Fetheringill-Jacobs atrium with AI from robotic arms to illumated coffee cups. The judges’ Awards showcased what happens when bold ideas meet real-world urgency. Taking first place, Dore-to-Dore by Sanjana Das and Niraj Girase reimagines crisis response as a living, adaptive system—using real-time data, location signals, and intelligent triage to connect needs with resources when every second counts. It is a powerful example of AI stepping into moments that matter most.
In second place, SchoolMe by Annika Coleman and Logan Pasquariello brings the same sense of immediacy to medical education. By transforming continuing education into short, adaptive, case-based experiences, the platform meets clinicians where they are—delivering personalized learning that evolves with their needs while generating insights that scale across organizations.
Third place went to PhysAI from the Neurdylab team, which addresses a foundational challenge in neuroimaging research: data quality. By embedding AI into the assessment of physiological signals, the platform strengthens the reliability of downstream analysis, enabling more confident discovery.
Together, these projects reflect a shared momentum—AI not as abstraction, but as infrastructure. Each team pushes beyond prototypes toward systems that integrate, adapt, and ultimately improve how we learn, respond, and discover.
PITCH: SPINS
Interpretation (MASI) Lab, developing substantial experience building software to visualize neuroimaging data. He began by contributing to
Open DIVE, an open-source brain visualization tool, implementing features such as coloring white matter streamlines based on their local orientation. Following that, he developed a longitudinal visualization pipeline for a clinical study of 134 ICU patients without traumatic brain injury, generating images for each patient showing brain volume at the start of the study and twelve months later along with their change in brain
volume through a color map. These experiences were foundational to the development of SPINS, where he applied those visualization skills to build a scalable pipeline for rendering brain regions across multiple atlases.
Learn more: https://masilab.github.io/SPINS/
PITCH: LANTERN
team is combining expertise in computational neuroscience,
educational research, and artificial intelligence to help K-8 teachers understand why students struggle with reading. Using
network analysis, artificial intelligence, and machine learning,
we analyze audio recordings of children retelling stories to
provide actionable insights that support early intervention and
personalized instruction.
PITCH: RESILIO
Researcher at the Institute for Software Integrated Systems
working with Dr. Abhishek Dubey, working on optimization
and decision-making for EV-grid coordination and V2B
energy systems, with a focus on online control under
uncertainty and multi-agent coordination. Fangqi’s work
focuses on developing AI and reinforcement learning methods
for intelligent decision-making in mobile IoT, cyber-physical
systems, and vehicle-to-grid (V2G) applications.
PITCH: Edge AI Enabled Ultra Low Field Body MR
Jonathan Martin (PhD), Charlotte Sappo (PhD), John Gore (PhD), Yashwant Kurmi (PhD), Jonathan Soslow (MD), Bruno Lima (MD/PhD) Ideas in Conversation
From Data to Gallery
What began as cutting-edge brain mapping has already crossed into something more—award-winning visual art now displayed alongside a UC Berkeley 7T scanner. In this moment, VALIANT’s work reminds us that discovery doesn’t just inform science—it reshapes how we see it.VALIANT Ventures
- Mingxing Rao presented the Deeper Dive on “Targeting Model Inversion from Generative Models.”
- Building off the new Vanderbilt-Leeds partnership, Dr. Irvin Teh visited VALIANT to discuss cardiac diffusion MRI.
- Building off the new Vanderbilt-Liverpool partnership, Dr. Yalin Zheng visted VALIANT to discuss AI in opthamology.
Alchemists’ Corner
- Effect of recoils on soft-drop-groomed observables in γ-tagged jets in a multistage approach
- Sloan Digital Sky Survey. V. Pioneering Panoptic Spectroscopy
- Protocadherin 20 Is a POU Class 2 Homeobox 3 Target Gene Required for Proper Tuft Cell Microvillus Organization
- Phrase-level emotional salience modulates neural substrates of situation model building in developing readers
- Longitudinal Changes in White Matter Hypointensities in Recurrent Late-Life Depression
- Bridging Histology and Tractography: First In Vivo Visualization of Short-Range Prefrontal Connections Informed by Primate Tract-Tracing
- Rapid trapping and label-free optical characterization of single nanoscale extracellular vesicles and nanoparticles in solution
- Giant Outer Transiting Exoplanet Mass (GOTa EM) survey: VII. TOI-6041: A multi-planet system including a warm Neptune exhibiting strong transit-timing variations
- Plasma lipidomics, choline metabolites, and metabolic-associated steatotic liver disease (MASLD): A Coronary Artery Risk Development in Young Adults (CARDIA) study
- Non-Small Cell Lung Cancer, Version 4.2026, NCCN Clinical Practice Guidelines In Oncology
- Multimodal Cardiac Imaging of Immune-Mediated Complications Post Heart Transplantation: A Contemporary Review
- Using diffusion MRI to relate hippocampal subfield microstructure to delayed verbal memory in cognitively intact individuals at genetic risk for developing Alzheimer’s disease
- Pathobiont-triggered induction of goblet cell response drives regional susceptibility to inflammatory bowel disease
- Did you know? State-of-the-art preprocessing diffusion MRI data can improve tractography
- SPYGLASS. VII-A. The Demographics and Ages of Small Nearby Young Associations
- The Importance of Standardizing Spectra in the Era of Large Spectroscopic Surveys: A Case Study of M Dwarfs in SDSS-V
- Foundation Models for Volumetric Medical Imaging: Opportunities, Challenges, and Future Directions
- Policy Search, Retrieval, and Composition via Task Similarity in Collaborative Agentic Systems
- Neural-Augmented Kelvinlet for Real-Time Soft Tissue Deformation Modeling
- Barriers and facilitators to implementing a shared decision-making tool for anticoagulant-related drug–drug interactions: a qualitative study across three academic medical centres in the USA
- An Oasis in the Brown Dwarf Desert: Confirmation of Two Low-mass Transiting Brown Dwarfs Discovered by TESS
- TOI-1080 b: a temperate, rocky planet orbiting a quiet M4V host
- Health system patterns of imaging and fluid biomarker testing in the era of anti-amyloid therapies
- Pulmonary artery size on computed tomography associates with mean pulmonary artery pressure and mortality
- Towards Transcranial Functional Ultrasound Imaging Through the Adult Skull
- Graph2Video: Leveraging Video Models to Model Dynamic Graph Evolution
- Mapping the Distant and Metal-poor Milky Way with SDSS-V
- Blinded by the Beam: A Unified Real-Time Defense Against Laser-Based Attacks on Navigational Perception of Autonomous Vehicles
- The Two-zone Temperature Distribution Model: Inferences on the Structure and Composition of Dusty Protoplanetary Disks
Resources
