Arjun Oberoi and Daniel Kim shared something in common: When their grandparents began losing their eyesight, they had to change the way they interacted with the world and needed assistance in ways they didn’t before. The two Stanford University students decided to help by building Argus, a two-part device that clips onto eyeglasses to help people with low vision navigate the world around them. With a camera and compute module, it responds to voice commands and answers questions about a user’s environment. Oberoi and Kim say Argus does three main things: identifies faces, navigates spaces and recognizes things like pills and other everyday objects. 

The device uses Wi-R technology and a combination of edge and cloud AI, including Azure AI Speech for voice recognition and Azure OpenAI Service for image analysis. 

Kim’s grandmother, who has a second-degree black belt in taekwondo and grew up amid the Korean War, was a big reason they focused their invention on vision loss. Watching someone with so many capabilities face new daily challenges was hard to watch, he says.  

Oberoi drew similar inspiration from his family as well. He says he hopes to make relearning new ways to do daily tasks a little easier for his grandpa “and the hundreds of millions like him.” 

The pair hope to do clinical trials with Argus to get U.S. Food and Drug Administration approval. 

Kim says that growing up, he always envisioned himself becoming a doctor to help others. But he says their success with Argus — as well as meeting people working to build their own products and companies — has given him new ideas for how “you can actually scale helping people.” 

Oberoi says competing in Imagine Cup has provided a great sense of community, giving him and Kim the opportunity to connect and learn from Microsoft mentors and many other student developers.  

Their main goal now is to get their technology into the hands of the many people who need it, he says, “to actually help real people in real life.” 

Signvrse

Branice Kazira of African Leadership University found herself tasked with teaching children who are deaf or hard of hearing at a robotics boot camp a few years ago without a sign language interpreter. 

Organizers told her to just “try to do something to keep them engaged.”  

“That experience was that ‘a-ha moment,’” Kazira says. “This is something some people actually face every day … they find themselves in a situation where they are unable to communicate.”  

She joined forces with Daniel Philip of United States International University, Anthony Marugu of Machakos University and Gheida Abdala Omar of Technical University of Mombasa to build an application for mobile and desktop that can translate either spoken or written words into Kenyan Sign Language. Signvrse’s app uses 3D avatar technology and Microsoft Azure AI Speech to power a virtual sign language interpreter.   

“Think of it like having your own avatar being able to do interpretation for you whenever you need it,” Kazira says. “It’s like moving around with a mini interpreter version of yourself.” 

The app has deep meaning for Philip, who is deaf. He says that growing up in rural Kenya, he felt ostracized because many people couldn’t communicate with him. Some even thought he was “cursed.” 

“They said things that cut deep,” Philip wrote in an email. “Adults mocked me. Kids laughed at me.” 

But he says he knows that his childhood is a story “shared by thousands of other disabled children who are still waiting for someone to believe in them, to give them access and to embrace their potential.” 

Signvrse is already in six schools, the team says, and they hope to get their technology into schools across Africa. That means they will have to expand their product to include more languages — a tall order, because they had to create their own motion-capture data set just to train Signvrse on Kenyan Sign Language. 

Marugu says there are very few motion-capture data sets of other African sign languages, so the team is creating them to be used in training AI models. They also hope Signvrse will one day be able to translate sign language into written or spoken words.  

“I believe that the technology is the next big thing in the world,” Omar says. “We are working on ensuring that communication is no longer a barrier for young people with hearing impairment.” 

HairMatch

 As a young child, Alycea Adams struggled to love her type 4A curls.  

“Growing up with my natural hair was something that I kind of dreaded in the mornings, waking up early and trying to tussle with my mom as she tried to get my hair together as we went out the door for school,” the University of North Carolina – Chapel Hill student recalls. 

“I dreaded picture day because my hair always looked not put together or just in a disarray because my hair just wouldn’t cooperate,” she says. “Once I started finding the products, I would try it, it would go to the shelf of shame underneath my cabinet, and I was just constantly wasting money.”  

She and her teammate Matt Steele, a Georgia Institute of Technology student, built HairMatch, an app that scans your hair and analyzes its type and porosity — then makes product recommendations and offers step-by-step hairstyle guides and expert advice. HairMatch uses OpenAI GPT-4o on Azure AI Foundry to process image data and enhance hair analysis. Steele says it’s one of the few AI-powered solutions in the natural hair space and already has users in 45 countries. 

“Hair is a universal experience, no matter where you may be in the world,” Adams says. “We’re trying to solve the frustration of bad hair days.” 

 When she met Steele at an Atlanta internship in 2023, the two became fast friends and spent their Georgia summer bonding over their shared athleticism — they’re both former Division 1 athletes — and Steele’s complete and utter lack of knowledge about haircare.  

“I was very misguided with hair products before Alycea,” he says, explaining that the chlorine from his swim practices had fried his normally straight hair. “I actually shaved my head after I stopped swimming. When I first met her, I had a shaved head.” 

Adams said she initially pitched the idea for HairMatch to Steele “as a joke” that quickly became a viable project. 

“So when Alycea had this problem finding hair products, we sat down, we tried to figure out, ‘OK, is this even something that’s technically possible today?’ And the answer is yes,” Steele says. “And then from there we went, ‘OK, well, it’s possible to make, it can impact a lot of people. We should build it.’” 

Now an entrepreneur and social media influencer, Adams says she has “the confidence to really own the curls and wear it as a crown.” She also hopes to be a good role model for other young Black and African American women in tech. 

“Representation truly does matter,” she says. “To see a role model go before you kind of lays down the path to give you the confidence that you can do it, too.”

This story was published on May 12, 2025, and updated on May 19, 2025, to reflect the winner. Lead image: Three student teams – HairMatch, Argus and Signvrse – made it to the final round of Microsoft’s annual global technology competition, Imagine Cup. (Illustration by Nicolas Smud /Makeshift)