Columbia Mailman Student Takes Top Honors in NVIDIA Hackathon

Columbia Mailman student Riya Kalra took first place in the December NVIDIA AI Summit Hackathon in New York City, beating a talented field of engineers and computer scientists. Created over a day and a half using one of NVIDIA’s most advanced chips, her prize-winning RxGuardian app was designed to help people stay on track with their medications and avoid dangerous interactions.

Kalra, who will earn an MS in biostatistics this spring, was inspired by her grandmother in India, who has poor vision and struggles with what pills to take. “At 4:30 am, my phone rang. It was my grandmother, 9,000 miles away in Gujarat, India. She had dropped her pills on the table. They all looked the same. She didn’t know which ones to take,” Kalra explains. “That moment became the 9,000-mile problem: when someone we love needs help, but distance makes it impossible. I built RxGuardian to solve that problem.”

How did you first get interested in data science and public health?

Kalra: I grew up in the Bay Area, near the heart of the software engineering world. At first, I thought coding was the only viable career path. I enjoyed computer science because it involves problem-solving and building products, but I was also interested in medicine. Then I found an intersection of both interests in bioinformatics, which involves using algorithms to analyze biological data. When I graduated from college, the AI boom was underway, so I decided to ground myself in the theory behind AI and data science. Instead of just using ChatGPT, I wanted to learn how to build my own AI. That’s why I came to Mailman to study biostatistics.

Tell me about the hackathon. How did you get involved?

Kalra: I took part in one hackathon previously. One of my teammates told me about the NVIDIA hackathon, so I applied. They gave participants access to a few AI models and had us run them on their new Grace Blackwell GB10 Superchips. We had free rein to build whatever we wanted. Unlike the previous hackathon, I was solo this time. I didn’t know what I wanted to build at first, but then I realized I could leverage image processing capabilities in the models, which, along with my grandmother’s medication needs, led to the idea for RxGuardian. My data science training at Mailman was super helpful. Understanding the theory behind AI helped me avoid creating AI "slop." I wasn’t just blindly vibe-coding. I finished coding the app by the end of the first day, leaving the second day for my presentation.

How does the app work?

Kalra: You can use the app to take a picture of your prescription bottles or your written prescription, and the model pulls out the names of the prescriptions and their dosages. You can also take a picture of the pills in the container or in your hand, and it tells you what the pills are, if you’re taking the right amount, and if there are any dangerous interactions. For someone like my grandmother, it’s not enough to have an alarm that goes off that says, “take your pills.”

What was your grandmother’s reaction? How did the judges respond?

Kalra: My grandmother was so happy. She’s my biggest cheerleader. I actually didn’t expect to win. There were so many intelligent and creative people there. The story of my grandmother made a difference. It resonated. When I presented my pitch, I used a picture of her. People came up to me afterward and told me they loved the story I told. It felt relatable and human.  

What’s next for you? Is RxGuardian available to use?

Kalra: Not yet. It’s more of a proof-of-concept. I’m not planning to turn it into a product, which would need help from the legal and medical compliance experts. When someone’s health is at stake, everything has to be highly reliable. Right now I’m focused on my job. I recently joined Tennr as a forward-deployed engineer. I work with AI models. We help with insurance qualifications and eligibility. I talk to customers every day. We figure out what they need, then I code it up. I love what I'm doing. It’s super cool.