When Ranjan Das began his PhD in mechanical engineering at UC Merced, he found himself in new territory. His background was in tribology and machine design — drilling bits, wear-and-tear, the mechanics of contact. But his doctoral research would focus on something far more human: biomechanics and human gait analysis.
“It was a new ven
ture. I didn’t have a background in human biomechanics, so I used AI to learn about human movements, how muscles contribute to walking, and how ground forces affect the body,” Ranjan recalls.
AI became his bridge into this interdisciplinary field. Biomechanics demands knowledge of mechanics, electronics, computing, even materials science. Ranjan found that conversational AI platforms could connect concepts across domains much faster than a stack of scattered search results.
He describes the difference like this:
“With Google, you get thousands of resources, but they’re not aligned to your need. It’s like asking a crowded classroom of experts for advice — everyone talks at once, and you have to sort through it. But with AI, it’s like asking a professor who knows you and your goals. The answer is specific, directed, and easier to apply.”
Building Better Prosthetics, Computationally
Ranjan’s research aims toward a computational framework for prosthetics and exoskeleton design. Instead of repeatedly testing devices on patients — a process that can be physically taxing for volunteers — he envisions modeling movement first, attaching those models to virtual prosthetic designs, and only later moving into human trials.
“Our vision is to have this computational framework ready so that all the iterations can be done beforehand,” he explains. “If we can model a person’s movement properly in all aspects, then we can optimize prosthetic devices before giving them to users.”
Teaching with AI: The Virtual TA
Beyond research, Ranjan has been experimenting with AI in the classroom. As a TA, he built a custom GPT chatbottrained on course materials — lecture slides, lab manuals, and textbooks.
“Instead of searching the web and getting lost, students could ask this chatbot and get answers based only on our course resources,” he says.
The chatbot was designed to act less like a solution-provider and more like a virtual teaching assistant: guiding students, offering examples, and even throwing in friendly quizzes mid-session. Students could ask “naive” questions without fear of judgment, something Ranjan noticed was missing in many in-person classrooms.
“The goal wasn’t to replace TAs,” he stresses. “But this could be an addition — something available 24/7 outside office hours, nudging students toward the right direction.”
A Philosophy of AI Use
Ranjan is careful to distinguish between using AI as a shortcut and using it as a tool for deeper learning.
“I tell my students, don’t use AI in such a way that after you graduate, someone asks you a question and you have to run back to AI for the answer. Use it so that when you graduate, you have the solution.”
For himself, that philosophy means using ChatGPT to understand concepts and debug code, while turning to Gemini for finding resources and the latest papers. He experiments with prompts, asking the same question in different ways until he gets clarity — a habit he calls his own form of “prompt engineering.”
Looking Ahead
From modeling human gait for prosthetic design to building chatbots that help peers learn, Ranjan sees AI not as a crutch but as an accelerator. The key, he says, is awareness and guidance: teaching people how to use AI to amplify their learning rather than replace it.
“AI should act like a virtual TA,” he says. “Not giving you the solution, but guiding you to it.”
Would you like to tell us how you use AI tools to accelerate your research? Please reach out - email sghosh@ucmerced.edu
