Name: Briana Frank, IBM Cloud
Uploaded: 2025-03-21T18:07:00.000Z
Duration: 6 min 9 s

Clips
More from NVIDIA GTC 2025

Briana Frank

VP of Product

IBM Cloud

play_circle_outline AI use case for personal life

play_circle_outline Training AI model with running shoe and data

Info
Transcript

Briana Frank, IBM Cloud

Briana Frank

VP of Product IBM Cloud

Exploring the Role of AI in Personalized Athletic Training with IBM at NVIDIA GTC 2025

In this engaging discussion at NVIDIA GTC 2025, we join Briana Frank, a product manager at IBM Cloud. Frank shares insights from a personal project involving IBM's Red Hat DeepSeek-R1 InstructLabs, where they harness AI technology to enhance marathon training. This episode, hosted by Savannah Peterson of theCUBE Research, offers a unique intersection of technology and athletics, demonstrating how AI can be applied to personal lifestyle choices.

Frank brings th... Read more

explore Keep Exploring

What did the person recently do with InstructLabs for their personal life as a lifelong runner and product manager at IBM? add

What is the speaker's profession and project they are working on at IBM related to privacy and technology? add

bolt Powered by CUBE AI

Briana Frank, IBM Cloud

Savannah Peterson

>> Good afternoon, Nerd Fam, and welcome back to San Jose, California. We're here at NVIDIA GTC, perhaps the hottest and most hyped tech show of the year. I am very excited to be coming to you from the IBM booth today with an interesting story you might not be expecting from a conference very solely focused on compute. I'd like to welcome Brianna from IBM. Thank you so much for taking the time.

>> Thank you for having me.

Savannah Peterson

>> So I love this

>> because A, you were not expecting to be standing here talking to me, but you have a really interesting use case for AI that is truly about humans. We talk a lot about it in the tech world. We talk a lot about it in the enterprise, but the cool parts of AI are when they really touch our human lives. So, tell us what you recently did with InstructLabs for your personal life.

>> Okay, absolutely. So I am a lifelong runner and I also love technology. I'm a product manager at IBM and one of the things we're working on is Red Hat AI InstructLabs. So it allows you to take your private data and align it to a large language model to help with the accuracy, but also make sure that your data's private. What I decided to do, because I'm a runner, is I wanted to train my model on the running shoes I have in my closet. Now, there's a lot of advancements in material science, so there's all kinds of really neat shoe functions and there's foams and carbon plates and all kinds of bells and whistles.

Savannah Peterson

>> First of all, how many different types of running shoes do you have?

>> I have four in my running rotation, so don't judge me, but actually-

Savannah Peterson

>> No judgment. I was curious. It's relevant for the data we're talking about here.

>> Exactly. And it actually prevents injury

>> and the shoes last longer if you rotate them because the foams need to rest. So what I did is I trained my model on these specific shoes, but I also use my own running data that's private that I don't necessarily want to share with the world.

>> Was that through a fitness app or how did you do that?

>> With the Garmin app.

Savannah Peterson

>> So it records all my runs and then I also record what shoes I use for each run. And then so that records things like your ground contact time, your stride length, your cadence, so I can see which shoe I'm running the best in, which shoe I'm running fastest in. And then I can ask my model once I've trained it to say, "All right, I'm going on a long run, which shoe should I wear? " Or, "Hey, I have a recovery run," or, "It's raining, I'm going to be running on the treadmill. What shoe should I wear?"

Savannah Peterson

>> So how many runs of data did it take for you to be able to train these models?

>> It really didn't take that many. It took a few months. I think I took a few, maybe two to three months worth of data. And the way that the product works is you arrange your data in what we call a taxonomy. So you load that data into the taxonomy and then you create three question-answer pairs. So I might say, "How much does the Adidas Boston XII weigh? " And I give the answer. And then what the product does is it allows you to take those three question-answer pairs and create synthetic data, so it creates a thousand different question-answer pairs. So it saves a lot of time. You don't have to enter all that information yourself. And then the model is trained on that synthetic data.

Savannah Peterson

>> Okay. So were you surprised by anything? Did you think you trained differently in any of the shoes than you actually did?

>> It was interesting. There was one pair that I didn't realize I was running a little faster in that pair, so it was good.

Savannah Peterson

>> Ooh, so was that the pair you raced in?

>> I actually used, I pulled out a new pair for the race and the race was successful. I wanted to see if I could shave off a few minutes of my time and I was able to qualify for Boston my first try of the marathon distance. I was pretty happy.

Savannah Peterson

>> Oh my gosh, congratulations. - Thanks.

>> That is a really huge deal. - It was awesome.

>> And you just completed this task two days ago, correct?

Savannah Peterson

>> Two days ago. So I'm hobbling around a little bit today.

Savannah Peterson

>> Hobbling around, but hopefully still reveling in a little

>> bit of that accomplishment and glory?

Savannah Peterson

>> I'm excited. Yeah, it was definitely one of those things

>> that it's tough and if you can overcome and train, it's pretty fun. It's pretty fun.

>> It is pretty fun.

Savannah Peterson

>> So, how are you going to use these insights to help you train for Boston?

>> Yeah, I think that the more you start getting into this, the more complicated you want to make the model. And it's like, "Oh, how could I add nutrition into this? " And like, "Oh, how much electrolytes am I taking in? " Or, "How many gels should I be using and training?"

>> Or my sleep the night before even.

Savannah Peterson

>> Exactly. So when you start

>> to layer in all this health data, there's just so many possibilities.

>> There are a lot of possibilities. That is really exciting.

Savannah Peterson

>> Brianna, for folks who might be interested in figuring out how to do this themselves, what's your advice?

>> So we are in closed beta right now, but we will, this will be open to the public within about a month. And so go to IBM.com or Redhat.com and take it for a spin.

Savannah Peterson

>> Awesome. Well, we'll be sure and help pipe that as well. Fellow runner over here though, I'm not quite sure I'm a marathoner, although you're kind of-

>> Well, you could be. - ... inspiring me.

Savannah Peterson

>> You could be. - I know, I know. At

Savannah Peterson

>> this point I was always thinking half.

>> I was more of a sprinter back in the day,

>> but I really admire it. And this is so reflective of the culture of innovation within IBM. It's not just about talking about the tech, it's about implementing the tech. Is there any other tech you use for personal use here at IBM?

>> Well, we are in IBM cloud, so any type of app

>> that you want to build, we have all kinds of serverless technology and Kubernetes and logging and monitoring. So pretty much any app that you want to build, we have a good platform for it.

Savannah Peterson

>> Oh, well, I can't wait to see what people inspired by your exciting marathon story end up doing. InstructLabs Open Source, such an exciting project and such a fun use case. Brianna, thank you so much for taking the time.

>> Well, thanks for having me. - Yeah, absolutely.

Savannah Peterson

>> I hope all of you runners are feeling inspired and learning how your data could be powering your next marathon and helping you find the right shoes for the occasion. We're here at NVIDIA GTC in San Jose, California. My name's Savannah Peterson. You're watching The Cube, the leading source for enterprise tech news.