Every morning at a nuclear power plant, a room of experts spends critical minutes deciding which repairs come first.
In a medical clinic, a patient with diabetes and high blood pressure leaves with prescriptions from two doctors who never consulted with one another on a generic treatment plan that rarely considers the patient's preferences.
In a university office, a staff member sits down with dozens of scholarship accounts, hundreds of students and only limited time to match them.
Anahita Khojandi sees the same challenge in all three situations. She spent her career building systems that help people make faster and better decisions without taking the final call out of human hands.
"I was doing AI before AI was cool," said Khojandi, who joined the U of A in August as the Palmer Terrell Family Endowed Chair and a professor in the Department of Industrial Engineering.
She was recently recruited as an Arkansas Research Alliance Scholar, a program that matches research leaders with efforts that help the state.
Khojandi came from the University of Tennessee with a cluster of academic efforts, including a new U.S. Department of Energy award to develop an AI-powered decision support tool to assist human experts in the time-consuming task of prioritizing work orders in nuclear power plants.
Additionally, she has developed a treatment framework for medical clinics that ensures a patient's choice is included. And a matchmaking engine for scholarship offices to ensure universities support students and continue serving as engines of economic growth and prosperity.
Her discipline ties these efforts together. Industrial engineering has always been about systems and the tools to improve them, she said, and those tools have gotten better lately.
"A system can be a human body, a system can be a watershed, a system can be a nuclear power plant, a system can be a department that is allocating scholarships."
She chooses which systems to focus on based on what she cares about.
"I want to leave it better than I found it. I think this is a way for me to give back with my craft," she said.
"I want to empower humans to use AI as a tool and be able to do better."
Nuclear Decisions
The nuclear plant is the newest of the three efforts.
Every day, according to the plant operators her team consulted, a cross-functional group of experts spends about 45 minutes deciding which of the outstanding maintenance work orders come first. It might be a leak or a noisy pipe. The team weighs each task against everything else happening in the plant to prioritize.
Khojandi's project, funded by the Department of Energy and launched with a kickoff meeting this fall, will build a tool called O&Mgpt, short for operations and management, that reads work orders alongside plant context and probabilistic risk models and returns a ranked list to speed up task prioritization.
The three-year project joins Idaho National Laboratory, the Tennessee Valley Authority (or TVA), the University of Wisconsin-Madison and North Carolina Agricultural and Technical State University.
Success, she said, would give TVA a tool, freeing its staff's time to tackle other problems across the states it serves.
Asked for results, she laughed. "We just had our kickoff meeting an hour ago, so ask me in three years."
Personalized Medicine
Improving health and healthcare is the effort she has worked on the longest: cardiac devices, neurology and intensive care, integrating patient demographics, social and environmental factors, and, more recently, genetics.
The current project targets patients with multiple diseases, multiple doctors and little say in their treatment plan.
"They don't ask what you want," she said of the typical medical clinic visit. "They tell you what to do."
The university's 150 Forward Reaching Higher plan names metabolic health, including Arkansas's chronic disease crisis, as a research priority.
"I weigh differently than you. I am a woman, you're a man," she said. "Why are we getting the same pill and being told to take it morning and evening with the same meal?"
"Also, I rarely get asked about my preferences and my personal tolerance for certain side effects when I see my provider and get a prescription. Maybe I'm interested in trying some combination of lifestyle changes and meds first, but no one typically asks me about it."
Because treatment rarely takes the patient's preferences into account, the patient goes home, and the pills stay in the cabinet, she said.
"A treatment plan that's not followed is not a treatment plan."
Her model for patients who have both diabetes and hypertension does something she says the published literature does not. It treats the patient's own preferences as part of the mathematics.
Working with about 10 years of patient records from hospitals across Tennessee, obtained through the University of Tennessee Medical Center under data use agreements, the framework infers what patients want from what they have already done.
A patient who has been losing weight, with no illness or medication to explain it, has in effect announced an intention to use lifestyle as a treatment plan.
The model then matches the plan to the patient: medication alone for someone who wants that, and medication plus lifestyle change for someone willing to try, handling the two diseases together rather than one at a time.
"What I'm trying to do now is bring together a triad: an AI, a physician and a patient, all of them collaborating," she said. The physician brings empathy and judgment. The AI can lay out every alternative and its consequences over the years. The patient brings the agency.
"It is a framework, not yet a tool. A proposal for a prospective pilot is being written. The pilot would put the model in a clinic, where a doctor and a patient could see the options side by side. A conversational agent is the step after that.
"That's future work," she said.
Matching Students and Scholarships
The scholarship office is the problem she calls her own industry's dilemma.
Donors endow scholarships with conditions: this high school, that sport. Her approach pairs a large language model that reads the agreements and student records with an optimization engine that allocates funds so every dollar is awarded, every eligible student gets something, and it explains why.
"Press a button, we can solve this very difficult problem in a matter of seconds," she said.
Why Arkansas?
Khojandi earned her doctorate at the University of Pittsburgh, where her dissertation optimized follow-up care for patients with implanted cardiac devices, and was on the Tennessee faculty from 2014 until this summer.
She spent the 2024-2025 academic year as an AAAS Science & Technology Policy Fellow at the U.S. Department of Veterans Affairs.
She came to Arkansas for the next step, she said.
"As an academician, I've been publishing a lot. But can I really bring some of these ideas to clinic?"
The Arkansas Research Alliance appointment is the mechanism. She will be inducted into the ARA Academy of Scholars and Fellows this spring, and she serves as faculty mentor to the university's student chapter of the Institute for Operations Research and the Management Sciences.
"I came for an interview, and I kind of fell in love," she said.
She tells her students the model is never the point. "They may be building the model, but that model alone, nobody cares if it's not solving a problem that is really important."
Getting the work in the field, she said, takes a village: clinicians, nurses, plant operators, other engineers.
"We are delighted that Professor Khojandi has joined our faculty. Her expertise will provide a substantial expansion in research areas that will benefit our department, students and the field," said Chase Rainwater, head of the Department of Industrial Engineering.
Demanding Guardrails
She hears the arguments against AI from her students.
She finds her answer in the kitchen. She washes handmade plates by hand because they are delicate and need attention. Everything else goes in the dishwasher, and she puts the time saved back into the people she loves.
"To me, AI is yet another tool," she said. "Being against AI is like being against the dishwasher."
That does not make AI the physician, she cautions.
"I don't want to get a cancer diagnosis from AI. I want a human to hold my hand."
AI doesn't settle questions about consent and where the data goes. She asks those herself.
"I'm not saying all of the problems are solved. I'm saying there is promise."
What she wants from the students now in her classroom is that they stay in the room.
"I really want young people to embrace it, so that they can set it up correctly for all of us, instead of turning their back," she said. "They should be demanding those guardrails."
Ellen Brown contributed to this story.
About the College of Engineering: The University of Arkansas College of Engineering is the state's largest engineering school, enrolling 4,701 students, including 3,608 undergraduate students and 1,093 graduate students. The college has more than 150 faculty and researchers and nearly 200 staff. The college ranked 65th among public institutions in the 2026 U.S. News & World Report Best Graduate Engineering Schools rankings and had $66.6 million in new research awards in fiscal year 2026. Its strategic plan, Vision 2035, will dramatically increase the state's engineering graduates by 2035, an outcome that will add $3.9 billion annually to Arkansas' economy.
Contacts
Anahita Khojandi, professor of engineering
Department of Industrial Engineering
akhojandi@uark.edu
Christopher Spencer, associate director of marketing and communications
College of Engineering
479-575-4535, cjspence@uark.edu