鈥淎I should learn chemistry the way humans do 鈥� by seeing molecular structures, not just reading linear strings,鈥� Vyas says. 鈥淲hile large language models have shown promise for molecular property prediction, their reliance on representations like SMILES or SELFIES [textual representations] limits their ability to capture the rich structural cues chemists rely on.鈥�

According to Vyas, this work opens a new pathway for chemical predictions and molecular analysis, by creating an AI system that operates more intuitively.

A Challenging Vision

According to Vyas, one of the biggest challenges facing the field of artificial intelligence and computer vision is in shifting AI models from a textual to a visual understanding of chemical reactions.

鈥淢olecular images represent a very different data domain compared to the natural images or text that vision-language models are typically trained on.鈥� Vyas says, 鈥淢olecules contain highly specific structural relationships 鈥� bonding patterns, stereochemistry, and functional group arrangements 鈥� that are subtle yet crucial for property prediction.鈥�

Many VLM models have limited exposure to visual representations of scientific data, which makes training and adapting them to understand the nuances of molecules and their atomic structure a primary challenge.

Transforming How Scientists and AI See Chemistry

To address these challenges, Vyas and her research team developed a multi-modal data set for MolVision to refer to during its training. The data set pairs 2D diagrams with text-based descriptions on a variety of molecules and different atomic structures. Using this data set was crucial for training the MolVision VLM to integrate textual and visual information effectively. Using a LoRA (low rank adaptation) algorithm, the MolVision VLM is able to engage in billions of parameters worth of data enabling it to complete complex tasks such as molecular property prediction or chemical description without the cost of full retraining.

鈥淩ecent advances in vision鈥搇anguage models have transformed how AI understands the world, but most of that progress has focused on natural images and everyday language,鈥� says Yogesh Singh Rawat. 鈥淲ith MolVision, we鈥檙e bringing those same AI capabilities into chemistry 鈥� allowing models to reason about molecules visually, in ways that are much closer to how scientists actually think.鈥�

This work has the potential to transform drug discovery, the personalization of medicine, and even sustainable design and engineering. The research team also expects that 鈥渙ver the next few years we can expect this multimodal approach to reduce experimental screening burdens, support faster identification of promising drug candidates and materials, and offer more interpretable insights into structure-property relationships,鈥� Vyas says.

Vyas and her team here at 麻豆原创 plan to scale up the MolVision VLM project in terms of its data set and capabilities. The team plans to integrate the VLM model in chemistry with technologies using current AI neural networks and large molecular simulators to create hybrid systems that can combine symbolic, visual and physical reasoning.

Vyas will also participate in the upcoming where she will be presenting an exhibit on AI for chemistry and molecules. Those interested in viewing the exhibit can attend from 7:45 to 11:00 p.m. this Saturday on the 4th Floor.

As part of the Fall 2025 Senior Design Showcase, two research projects featured faculty mentorship that resulted in new ways for students and medical professionals to collaborate in health sciences and medical education.

Judges selected the two projects to be featured in the yearly showcase because of their quality and potential for real-world application.

The students received invaluable experience and the technologies developed are ready to be further refined by another incoming group of 麻豆原创 undergraduates.

Understanding Congenital Heart Defects Through Immersive Technology

If Nabeeha Vorajee were was in a clinical setting, she would need a mask, scalpel and medical scrubs to explore the anatomy of the heart. But she鈥檚 not a medical student and the heart she鈥檚 studying isn鈥檛 real thanks to new technology she helped develop.

痴辞谤补箩别别听颈蝉听补听蝉别苍颈辞谤听鲍颁贵听computer science聽major,聽and the heart is聽made of thousands of pixels residing in a聽simulated classroom accessible through virtual reality聽(VR).

Students wear a virtual reality headset to better understand CT scans of the heart. Using AI, the system categorizes the images for common heart defects. It also creates a 3D digital twin of the heart that students can interact with through the headset.

Vorajee鈥檚 team took the learning system a step further and applied it for medical education. They integrated AI learning and the ability to highlight, isolate and segment different parts of the heart. If users have questions about what they鈥檙e seeing, they can ask the AI data set without leaving the session.

Users can import cardiac CT scans and transform them into a 3D heart, which allows them to take the model and learn beyond the classroom, Vorajee says.

鈥淐omputers and VR headsets are a lot more easily available than an actual heart,鈥� she says. 鈥淚f a student wanted to study up, they鈥檒l be able to do so a lot easier with this program than they will be able to find a heart.鈥�

The project included guidance and mentorship by Laura Brattain, associate professor of medicine, and Matthew Gerber 鈥�98 鈥�00MS 鈥�05PhD and Richard Leinecker, associate lecturers of computer science.

Clinical Imaging VR

Before pursuing a degree in computer science, Arianna Ramirez Oquendo attended nursing school and used her experience to inform her group鈥檚 research project: a VR application to help medical students practice interpreting CT scans.

鈥淲hen I was in nursing, I realized that most of my practice identifying organ structures and reading CT scans or MRIs was restricted to only lab times,鈥� Oquendo recalls. 鈥淚 needed more than two or three hours a week to really understand what I’m looking at.鈥�

Oquendo鈥檚 group鈥檚 system simulates a clinical imaging office where students can compare normal and abnormal CT scans, identify anatomical structures and recognize medical implantable devices.

The app allows students to see and adjust up to 160 slices 鈥� or images 鈥� from each CT scan. After the tutorial and CT scan room, there is a final study room where teachers can implement multiple choice or 鈥渄rag and drop鈥� quizzes on specific scans.

Emily Bradshaw, an associate professor of medicine and project sponsor, says that this project has great potential to help students and supplement their education.

CT imaging can be a difficult radiologic discipline to master because the images are presented in three different planes, says Melissa Cowan, project sponsor and assistant director of instructional technology at the College of Medicine.

鈥淏y aligning the CT images to a 3D model of the patient鈥檚 skeleton and organs, the students have a visual reference as they scroll through the various views and isolate the key findings,鈥� she says. 鈥淭his app is another tool to help students learn how to read and interpret CT scans and apply their knowledge to other patient cases presented throughout the curriculum and in clinical settings.鈥�

A longer version of this story can be accessed on the .

Today, the 24-year-old is on the cusp of earning two graduate degrees simultaneously, having elevated 麻豆原创鈥檚 reputation as a community and society changemaker along the way with the AI for Nonprofits course she helped develop. These milestones aren鈥檛 just personal. They鈥檙e also serving her biggest passion: helping others.

That鈥檚 the thing about opportunity. Sometimes all you need is a chance, a dash of inspiration and someone who believes in you.

Charting Her Own Path

Dones, who was born in Cuba, was just 2 years old when her family immigrated to Miami. Growing up, she viewed attending college as a somewhat unattainable goal.

鈥淚 remember discussions with my parents where they would say that college is for the wealthy, or that we’d cross that bridge when we get there,鈥� she says. 鈥淚 didn’t have anyone to rely on for mentorship because my parents didn’t know anyone that had been to college before, and all of our family was in Cuba.鈥�

In high school, she joined the International Baccalaureate program and learned about financial aid opportunities such as Bright Futures scholarships. A teacher encouraged her to seek out more scholarship opportunities from Miami-based nonprofits and foundations. That鈥檚 when she realized college was within her reach.

鈥淚 thought, 鈥業f everyone around me is doing this, what’s stopping me?鈥欌�� she says.

Dones connected with the Key Biscayne Community Foundation, a nonprofit organization that helps students achieve their educational goals like the ones of her own. With assistance from the foundation and other nonprofits, Dones ended up receiving $20,000 in scholarships to attend 麻豆原创.

The support not only sparked her academic career; it also inspired a calling to give back to others.

She pulled up others with her along the way as an undergrad, serving as a mentor for local high school students and impressed upon them the importance of internships. She realized how much she loved positively impacting them.

That鈥檚 when her own mentor, Josefina Rosario 鈥� assistant director for access and student support at 鈥� suggested enrolling in the public administration and nonprofit management dual-degree graduate program next. It was a decision that would further fuel her passion for helping others.

Through the program, housed in , Dones has engaged in real-world experiences that simulate what it鈥檚 like to work in the nonprofit and public sectors. Through service-learning, she evaluated and helped develop a strategic fundraising plan for a local nonprofit. She鈥檚 also created budgets for the City of Orlando and analyzed how both the city and Orange County Sheriff鈥檚 Office use social media. She attended her first professional conference.

Perhaps some of the most invaluable experience she has gained involves the industry鈥檚 intersection with artificial intelligence.

Leveraging AI for Community Impact

Through the 麻豆原创 Applied AI Innovation Initiative, Dones has been working with associate lecturer and initiative lead 鈥�11笔丑顿 in using generative AI tools to find innovative solutions for challenges in educational and behavioral interventions. During her time as a student fellow for the initiative, Dones has learned to harness the power of AI to create apps, websites, data dashboards, chat bots, online courses and even a board game that enhances financial management skills 鈥� all without having a technical background.

鈥淚 never thought I would be able to build an app, but now we can with the help of AI,鈥� she says. 鈥淲hen I saw that I can do something like that or create a website with the help of AI, I wanted to continue doing it. Nonprofit organizations tend to get left behind when there’s new technology out there. I imagined how AI can help a nonprofit with time or resource constraints, and that鈥檚 how my passion for learning about it started.鈥�

鈥淚 imagined how AI can help a nonprofit with time or resource constraints, and that鈥檚 how my passion for learning about it started.鈥� 鈥� Ketty Dones 鈥�23

Dones has played a significant role for the Applied AI Innovation Initiative in helping local nonprofits leverage AI to boost creativity and efficiency in advancing their missions. She contributed to creating an AI Impact Hub, an app that serves as a one-stop shop to connect nonprofits with resources and engage with other nonprofit leaders through a discussion forum. She鈥檚 also assisted in hosting workshops that explore how AI can be used as a tool to assist nonprofits with social media, fundraising, compliance and evaluation.

Most notably, Dones helped the team develop an in partnership with that covers topics ranging from grant writing, evaluation, storytelling and prompting with AI to ethical guardrails, accountability and security when using the technology, as well as best practices for integrating AI tools in day-to-day operations. The course is also used in the Innovation and Technical Assistance Program, which provides hands-on assistance to local nonprofits engaging with AI-powered tools to help strengthen their organizational capacity.

In fact, 26 Orange County nonprofits graduated from the program last month with an AI for Nonprofits certificate.

Full Circle and Ever Upward

If it helps bridge the gap between AI and nonprofit organizations, count Dones in.

鈥淚 always like to think of AI as a thought partner and ask how it can help elevate ideas, not create them,鈥� Dones says. 鈥淭hat鈥檚 why I think it’s so important for the public and nonprofit sectors to leverage this technology. Continuing education in the nonprofit and public administration sectors is always important, but especially now with learning how to integrate AI tools.鈥�

Although AI has served as her assistant in creating innovative solutions for others, it鈥檚 the combination of her grit and determination to succeed along with a passion for serving others that assists her in unlocking the potential not just within herself but also in the community.

“Ketty has impressed me since day one,鈥� Concha says. 鈥淏etween her contributions to the AI for Nonprofits course and developing impactful resources using AI and advanced prompting techniques, she is a prime example of a student without a technical background using AI for social good.”

For now, Dones will continue working with Concha and the Applied AI Innovation Initiative to further develop AI tools and solutions, working toward certifications and mentoring nonprofit partners along the way. Although she is considering pursuing a career in AI product management or sales down the road, it鈥檚 using her knowledge to set up nonprofits for success that she finds most rewarding.

After all, her accomplishments thus far and her bright future wouldn鈥檛 have been possible without some help from nonprofit organizations along the way.

鈥淚 always think about the Key Biscayne Community Foundation and other nonprofit organizations that helped me receive funding to go to school,鈥� she says. 鈥淚t鈥檚 kind of full circle for me now. I feel that I can give back to my local community in understanding not just the challenges nonprofits face but also how I can help them in the future.鈥�