Landri Domingue grew up being told she communicated differently. Today, she is teaching artificial intelligence to communicate better.
Domingue, a rising junior studying computer science, is working this summer in Dr. Yu Huang’s Mixed Intelligence Development for Programming Lab (MIND Lab) through Vanderbilt’s Institute for Software Integrated Systems. Supported by the National Science Foundation and the Computing Research Association’s REU program, she and her collaborators are modeling human brain signals captured in functional MRI scans to fine tune and guide how large language models generate code.
“By integrating human cognitive process workflows into LLMs, we can significantly improve how they generate code by better mimicking human processes,” Domingue said.
The question at the center of her work: What does it look like, neurologically, when a person writes code—and can an AI learn from that?
From robotics to radiosurgery to research
Domingue was ten years old when she joined her school’s robotics team and, as she puts it, “fell madly in love.” Born with a severe speech articulation disorder and dyslexia, she had grown up finding communication a constant struggle. Robotics changed the dynamic.
“My classmates saw beyond my learning disabilities to recognize my contributions,” Domingue said. “I was a natural problem solver and could devise both complex and simple solutions.”
That foundation led to her first machine learning project in the summer of 2022, building a stroke risk classification pipeline. The following summer, she became Mary Bird Perkins Cancer Center’s first-ever high school researcher, working under the Chief of Physics to analyze software-predicted versus actual treatment times for Gamma Knife Radiosurgery. That work was published at the American Association of Physicists in Medicine 2024 Annual Meeting and Exhibition—with Domingue as an author.
“It was rewarding to see my first research project get accepted at a national conference,” she said. “I was incredibly excited by the possibility that my work could have a real-world impact. I was hooked from then on.”
A cold email, a podium, and a next step
When Domingue arrived at Vanderbilt as a freshman, she did what she had learned to do in high school: she cold-emailed labs. Dr. Xinqiang Yan welcomed her into his lab, where she led a machine learning algorithm project for ultra-high-field MRI.
In October 2025—her sophomore year—she was selected as the sole undergraduate among more than 130 researchers to deliver an oral presentation at the 20th Annual Vanderbilt University Institute of Imaging Science Research Retreat.
“I’ll never forget presenting that first time—being on stage in front of a podium with a microphone, arm outstretched to the projector, enthusiastically explaining the algorithm I developed in front of a crowd of over 100 researchers,” Domingue said. “Growing up, I was taught to hate speaking in front of large crowds due to my speech articulation disorder. Instead, I found this experience to be invigorating.”
Another cold email followed—this one to Dr. Yu Huang. It led to her current role in the MIND Lab, where she, Huang, and PhD student Zachary Karas are bridging neuroscience and artificial intelligence in ways that weren’t possible just a few years ago.
Reading the brain to improve the machine
When people think of LLM fine-tuning, they tend to picture dense mathematics, transformer architectures, and attention blocks. Domingue didn’t expect that so much of her time would be spent looking at brain scans.
“I never thought I would spend so much time looking at brain scans and translating them,” she said.
The project draws from artificial intelligence, neuroscience, and software engineering—a combination she describes as a natural fit for the College of Connected Computing’s interdisciplinary spirit. Her days move between PyTorch, TensorFlow code writing, training deep learning architectures, brain image analysis, and literature review, punctuated by weekly one-on-ones with Karas and joint lab meetings with Huang.
The research topic itself crystallized during a pivotal lab meeting in which the team reviewed a previous publication and began trading ideas rapidly, each one building on the last.
“We started ideating and spitballing back and forth, excitedly building off each previous piece thrown out, until we got to it and immediately knew—yes, that’s our topic,” Domingue said.
Like many open research questions, the challenge—and the appeal—is the absence of a roadmap. There is no established playbook for how to translate neurological data into model improvements.
“If you already knew before you set out what you would uncover and precisely what steps you would use, then it would not truly be research,” Domingue said. “You get to map each step.”
An intertwining cord, not a race
Domingue is frank about where she sees this field heading. She does not believe AI will advance primarily by outpacing human intelligence—but by learning from it.
“I see the future of AI more as an intertwining cord with humanity than as a linear line racing against it,” she said.
She plans to continue the project into the fall semester with the goal of publication, and intends to build a career in AI and machine learning. Her personal philosophy, borrowed from her South Louisiana roots, keeps the momentum going even on slow days in the lab.
“Back home in Louisiana, we have a saying: ‘Laissez les bons temps rouler’—let the good times roll,” Domingue said. “So, when the alarm goes off in the morning, laissez les bons temps rouler.”
For students hesitant to reach out to a lab, she offers the same advice that opened every door in her own research career.
“Don’t be afraid to reach out first and dive in with curiosity,” Domingue said. “Being engaged and fascinated can open doors you didn’t even know existed.”