How Data Science Helped Build a Smarter Fitness App

Manikanta Sirumalla, a graduate student in UMBC’s Data Science program, shares the story behind Rep Track Pro — an AI-powered fitness app designed to bring workouts, nutrition, recovery, and progress tracking into one place.

In this conversation, he discusses how his own fitness journey inspired the idea, the challenges of building and launching an app as a solo founder, and how coursework in UMBC’s Data Science program helped shape the technology behind it. From machine learning models to real-world user feedback, this is a story about innovation, persistence, and building something meaningful from the ground up.

Learn more about UMBC’s Data Science graduate program:
https://professionalprograms.umbc.edu/data-science

Learn more about RepTrack Pro App: https://reptrackpro.org/

Transcript

Dennise Cardona  12:37:03  

Hey, thanks for tuning into this episode of UMBC Mic'd Up Podcast. I'm Dennise Cardona, your host, and it's wonderful to have you here. I am here today with a current student in our data science graduate program. It's so nice to have you here. Manikanta, thanks so much for being here, and I can't wait to talk to you about your experience with UMBC, but also this experience that you have building an app. So to start us off, can you first of all share a little bit about your background and what led you to the data science graduate program here at UMBC?

Manikanta Sirumalla  12:37:41  

Oh, that's a wonderful introduction. Cardona and myself. I am manikanta sirmala. I'm from India, so coming to my background, I'm a professional iOS developer who worked for three years in India, and like after that, I wanted to explore more opportunities in data science field. Why? Because, like, while, while my work, while I was doing my work, I've dealt with the different type of applications where data is a way the data played the major role. So I really wanted to go deeper into that one and building apps with the data. I think that that's where the mobile mobile future is going to be. Mobile App future is going to be. Then I started looking around the universities that could give me the better opportunities in exploring these things, and I found, luckily, I found UMBC and the BW tech and all, so I directly uploaded and finally I got into UMBC.

Dennise Cardona  12:38:32  

Yeah, great. I'm glad that you found yourself here. It's wonderful to have you as part of our community. You're welcome. So I hear that you've taken your experience beyond the classroom, and you built yourself something real, an app called rep Track Pro. Can you tell us a little bit about this 10 year personal fitness journey behind it, and also what was missing in the existing tools that made you say I want to build this app?

Manikanta Sirumalla  12:39:07  

Oh, god, that's a very long story. So I've been training like 10 years right now, and honestly, when I started, I was that guy who don't know what to do. I just stepped into the gym. I was skinny, and there was no tracking, no idea what I was doing. I feel like, very glad there are people who are bigger in size than me. I want to become like that, but I don't know how, and what's the path to get, to get to that one. So over the years, like, I got training, I got serious about it, and while I looking changes in my body in the mirror, then then the real motivation started, like, I want to take it to the next level. So like everyone else, I tried every fitness app out there, like, let's say My Fitness Pal for food heavy for logging workouts, and fit board for generation of the plants, and Apple Health, generally, Apple Health for the steps and all the health data. The problem that I noticed was that none of them talk to each other, you know, like, the calories in my fitness pal stays in. My Fitness Pal, the working while we are logging out that stays in heavy they do not communicate to each other. So if I want to get something, if I want to get progressive in my journey, they should be talking right? Let's say these are the calories that I ate today, and this is my workout. So what if I ate the same thing tomorrow? Is it going to the same so that kind of analysis should be done so that I can build up on myself. So that is the main reason that I have started looking this one. And the plans were so generic, the plans that I generated were so generic, and same workouts that pushed like a 20 year old, and the same workouts I guided my dad to, then the same workers for his profile. So that made me so this is the biggest gap that I could found. And by then, the idea was ready, but I'm not equipped with so much technical skills, okay? And then after, after some some sort of things, I don't have time. I was equipped with the work and all. And when I got here in the UMBC, then I slowly realized, like, now I've got some time for academic projects and working on my own projects.  Then where it all started? 

Dennise Cardona  12:41:09  

Can you take us back to the moment that the idea clicked? When did you realize that this could become a real product, not just a side project?

Manikanta Sirumalla  12:41:22  

Yeah, honestly, it didn't happen in one big lightning moment. It built over time. You know, first the shift happened when I was trying to plan my own own week, like the next week, from Saturday, from Monday to the next Sunday. So I had my work course in one app, same the same problem, one app for the another and the way that I logged in another app. So literally, copying all the numbers from different applications, and writing it down, and then planning my week accordingly. I felt that is ridiculous, so I'm doing all this work just to manage the tools which are supposed to work to me, and working for them, taking all the notes and all and preparing my week. Then the moment that really pushed me is like, is that when I came to the gym the UMBC in 25 If I saw a guy logging his workouts with a pen and paper, literally, he's logging every workout with a pen and paper. And I went to him straight and I asked him, like, like, even after so much of technical advancements, why are you doing this? Like, you could have used your phone and everything. And that answer from him is just blown my mind. Like he said, like, all the apps that I have tried, they are all garbage, like, they charge more than what they do exactly, he said me that sentence. So I was like, it's the same problem that I was going through. Like, there are, there are many people, like, there's a bigger opportunity in this thing, right? So that hit me, and exactly the fitness apps are supposed to help. Even some of them are serious. Don't trust the app. So these are the serious lifters, like me, who is having 10 years of experience, and the guy whom I talked to, we are the serious lifters, and we wanted to have some progress. We want to look up some some progress in that thing. So this is the, this is a main scenario that clicked in my head. Then I started like, how would it be like connecting every application with the diet food workers AI into it, and, let's say, AI coach now, now, now we have plenty of new features in that one too. I'll be explaining in the later part of the podcast. And being a master, being a Masters in data science students and iOS developer for three years of experience, and what could be the better platform to build that one and just distribute it to the world. So that is the main thing.

Dennise Cardona  12:43:28  

So you've built and launched this as a solo founder, from what I understand, what were the I would say, what were the biggest challenges that you face going from the idea to a Live app in the App Store?

Manikanta Sirumalla  12:43:42  

This took me around one and a half years to build this one, and it's a not, it's not the same application that what people are looking right now. It's very, completely different application that I built it for myself for that that works according to my my kind of thing and my experience. Then after getting through the different people, talking to the different people, like the guy in the gym that I recently said, and I realized the problem was real, and it's bigger than I expected. So it's not for me, if, if this could become a business model, this is going to help, like many people. So I shifted the gears. I wanted to take this to the much larger audiences that actually helps them, just not just not myself. So the biggest challenge, honestly, was that time like I started it in the first semester, it was easy just to figure it out and make the make all the documentation, all that was easy in the first semester. Then the biggest thing came. It's my second semester, and I got enrolled in three core subjects. Then I have to manage three projects. I have to manage this app. And there's so much of time constraints and the technical thing that's happening around there were some sleepless nights that I spent over different different projects, like, there are multiple times, multiple months that I've spent without sleep. You know, there's just sleeping for three to four hours, doing all the things, all the things and all So, and this is one of the challenge and the second challenge is the iterative feedback of the Apple ecosystem, like whenever, when I, when I given an app for the submission, they literally canceled me, rejected me for like, five times in a row and again. Then that's an iterative process, right? They cancel, they reject me, then I have to recheck and submit it. Then they will check me. And that was a mess. And finally, I was not expecting the mail to come and again, I was expecting the same rejected, rejected, rejected, rejected mail in my in my mailbox. And finally, one day that there is a major Mail Call like successfully deployed in the app store. And that's when all my all my work, got a real thing from the apple. And the next thing was like, the App Store reviews, let's say, being a first version of the app. There will be glitches. Of course, I am the solo designer, developer and everything by myself. There will be glitches. There. There are some errors. There are people who are complaining about the crashes they are. There are people who are giving me feedback, which is, this is nice. This is not nice. And at times, there are people who are who are working out at 2am in the morning, and they found it glitch. Then they mailed me about that one. Then I got to wake up and check on that one, and then then submit it to the next version. So these are all the things that that made me like the biggest challenges that I could face here.

Dennise Cardona  12:46:18  

You know, I think it's like with life, if you don't have those challenges, then how are you going to grow, right? And so the same thing with technology, everybody's got to expect that new technology is going to have glitches. I mean, every other day I'm getting a, you know, a new a new patch in my iOS software. And, you know, this thing has to be the firmware has to be updated. So it happens. And I think, to me, that signals that there's growth in the technology I'm using, and that they're paying attention to it. So I think that's a good thing. And, yeah, it's just part of the challenge. It's, it's like anything, any project you put out there into the world, anything, yeah, you're going to have that feedback. And I think the true measure. Of the character of you as the founder and the creator is that you're able to accept that, that feedback and do something positive with it. So kudos.

Manikanta Sirumalla  12:47:12  

Good for you. Maybe all of those, those sleeplessness like, get me the state that I'm here right now doing podcasts with you.

Dennise Cardona  12:47:20  

That's right. That's right now, your app uses AI for workouts and nutrition planning. How did you approach building that system in a way that's both smart and practical for everyday users?

Manikanta Sirumalla  12:47:36  

So the core idea of implementing AI in the application is that, like most so called, so called a fitness app, send you the info that you provided to the chat, GPT or any LLM that they that they have, and as they're dumb as the AI gives the response Exactly. That's what happening in the in the present AI fitness apps. So that could be break of in the three ways. Like one, AI makes up exercises that don't exist. Two, it ignores the what equipment that you have. Let's say I'm having access to full gym in UMBC, the some of the people don't have, they will be having just a dumbbells and the basic stuff. So even, even, even they are getting the same kind of replace from the LLM who is having full gym equipment so that that that is that there's no point of having the AI in that one, right? And other thing is nutrition advice drifts away from the Real Sports Science, like in my application, the every nutritional value, or the every nutrition that I've, that I generate, that I gave to a is completely based on the science, based, you know, science, science research, back and out of it. So these are the things that I found. And I don't want to repeat that. I don't want to be one of the application that does all these things, you know. Then I started building rep Track Pro, like imposing all the AI can a content into that one. I call it like the deterministic architecture first, and AI fills in the content. You know. Let me explain like this is not a math in a or a calories, protein, macros, protein recovery, all are computed, but the evidence by the evidence based formulas, same person, same numbers every time. So we have validated this to the DEXA scans. Also, like I did, I did DEXA scan by myself, and I validated the metrics and the values that that my app is going to give you will be matching the same offline DEXA clinical scan or not. Then, then I made some tweaks into that one, and finally I got the sweet spot, like I'm getting 90% accurate to DEXA. So I think for a mobile application, just to know the trends. I'm not saying that this will replace the DEXA scan. I'm just saying to follow the trends. No Anything is fine to if you find progress in something, this is other thing. And the training structures are, like, completely not filled by AI that the I built a determinate, deterministic architecture that we will be having warm ups, main exercises and deloaders. As you are more into fitness, you know about this terminology. So the only thing AI does is that it takes from all the exercises that I have in database. In my database, I'll be having like, 2000 plus exercises. Ai goes through that one, and it suggests you the best, best exercises according to your goal. Let's say I'm bulking up right now. Let's say I'm bulking up. I want more intensity, so I will be having more compound movements. For the people who wanted to cut down, they'll be having more hypertrophy, so more reps and less weight. They'll be the a will be suggesting that kind of exercises to the people who are there who wanted to, like, cut down in the weight or in the fat loss also, like, I also build in this thing in such a way that you'll be having every YouTube video for 2000 exercises in the database. Like, I don't want users to go out of my app and go, you go search for the exercise and all. I just wanted to be there. I just wanted to learn from the YouTube or any kind of exercises or videos that I put in there with the variations, tips, tricks and common mistakes. So that's what it all happened.

Dennise Cardona  12:50:56  

Oh, I love that. That's a really great because that that happens, you have to go out, you have to go to YouTube, you have to figure it out. Search for it, for the particular exercise, you find a million different things that are not relevant, so now all of a sudden they're off your app. So that's a really smart one. Just sounds so smart, yeah?

Manikanta Sirumalla  12:51:14  

Also, like in the latter part of your question, is, like the practical experiences, like every day for the everyday users, it's a honest that it's good to say I don't know, rather than faking it right for any a or anything, rather than making it up, making it out, it's good to say I don't know, and just working exactly what you have, what the data you have, and what could what could you bring about? How could you add the value to the people's life? So that is a main thing, and it's boring. But if AI fails in, let's say if AI fails in, generating the workouts, I built an app in such a way that the deterministic architecture gives you the offline accelerators too. You know, sometimes we will not be having internet connection. And maybe while generating the plan, you have a difference in the in the connection wise. So if the, even if that has. Happens, you will be having an offline version that gives you the same output out of it. 

Dennise Cardona  12:52:08  

Ooh, that's that's that. That's very valuable. How did your course work in the UMBC data science program? Shape the way that you designed and built rep track, Pro? Are there any specific concepts or classes that really helped you to engage with this?

Manikanta Sirumalla  12:52:27  

Honestly, the class that changed everything that I think of right now is a machine learning fundamentals from masharash. Professor, masharash, I have to thank him a lot. Like in my second semester, I took that as my core subject, and while taking the class probably the single best ever decision that I've taken that in the in this, in this program. So before that, I had a plenty of printing narrow knowledge about the AI machine learning stuff and all, and I only know the chat bots and llms that we are using for the AI. That's the only, only thing that I know about the A so when I joined this class, like it'll be as he was going to the classes we in the like four to five weeks down in the classes. Then I thought, then I thought like, the the way is teaching, and all the regressors, classifiers and all the stuff in the machine learning, one day, something clicked in my mind. I wanted to explore it in my application, to in the in the real world application. So I directly went to the muscles, Professor muscle, asked him, like, sir, this is, this is what I am building, and I wanted to do this in certain way. So could you please help me out to finding out the right direction and right model to to to take it in a in a direction I'm not saying it's 100% correct, or it should be 100% correct, at least, to the say, to the direction that it could take me to get better values, you know, better better results. So that is the biggest decision I have taken, and he is very, very helpful for me in that case. And like, let's say, a few examples, like, in my application right now, I have few models, like regression models and classifiers. The regression models I'm using it for the weight prediction and the classifiers I'm using for the recovery models for, let's say, how good you are recovered from the yesterday workout to do two days workout. So these are all the coursework that I've learned from the master service class, and I implemented in my application is working successfully. Also in this semester, I enrolled in project management fundamentals from Auden Williams, and I have to thank him, like is so good, and he's explaining it so well. So let's say I'm a solo founder. Right from the development perspective, I know I have some physical experience, like the work experience for three years, and some technical experience coming to a and machine learning from matsura's class and all the all the learnings that I've had over the years from the nine years. And how about the management thing? Like, I think his class, like, it's amazing. I'm learning so much out of it, and I'm very excited to take a rep Track Pro in the direction that I've learned.

Dennise Cardona  12:54:48  

I love hearing that. Well, you recently won first place at the Cangialosi Business Innovation Competition. What did that experience teach you about telling your story and positioning your product?

Manikanta Sirumalla  12:55:03  

Yeah, winning CBS business innovation competition is a huge moment for me. But honestly, the experience taught me more than the prize money. Did you know The biggest lesson was, don't pitch what you can't defend. I think that's that's a line that I learned from that from the competition. Like early on I was leading off. Early on, I have my, as I said in the recent conversation, in the in the past, question, like I have offline mode in the in the project. So it sounded great on the paper, actually, but when I imagine judges are digging into it and asking some technical questions about that, when I was not so confident to answer that thing, and I thought of like taking it out completely and replacing it with a 90% X accurate, which I'm very much confident about the metrics and values because I had my my DEXA scan right now in the physical, physical thing. So I changed that one, and that helped me a lot. Actually. You know, the lessons were simple, like, judges aren't impressed by the feature list. Actually, they impressed. But how well can you present your your topic or your project and hold up under pressure? That's the thing. And the second biggest lesson was about framing like I used to think like pre revenue and being a solo founder of weaknesses, but later, when I found I pitched in such a way that built from zero and profitable from subscriber one is my headline, and that that changed completely.

Dennise Cardona  12:56:30  

Yeah. Well, the marketing I'm a marketing professional, so that, like, totally sits and resides in my heart, because, you know, you have to be able to market yourself. You really do. And that's a very catchy, catchy motto that you have there. Yeah, yeah. Now for someone listening who's more technical, what's one part of the apps, architecture or design that you are especially proud of?

Manikanta Sirumalla  12:56:58  

There are many features that I'm proud of, but the deterministic i. Architecture while splitting the AI exercises is the one thing that's that's no one did in the in the literature too. I have searched the literature also, then no one did that one basically, what does it means like for technical folks who is listening? This is the, this is a part they most proud of, to build an architecture that works in offline and online mode, and not giving a guesswork to the AI to what to guess exactly the exercises and all, it actually gets you through the deterministic architecture, through the architecture that I built, and it fills the exercises that what I had in my database, right? So that is the thing I'm very proud of, like in the recent applications, as I talked earlier to the recent applications have tried different, many different application many different applications, and directly giving my data to the chat GPT or some kind of llms, it just guesses the work and just gives some other random exercises that some equipments are not available in our zoom too. So that is the one thing I'm very proud of. And the second thing is, recently I've added for the females, the menstrual cycle, away training and that, and that's going that's you guys gonna love it like it adapts to you. It seemed like the flow application and other other menstrual cycle tracking applications, but they do their own exclusive work, right? But what's in our fitness app is that it suggests you, it adapts to you accordingly, like, when you feel like you are energized let's say, in a follicular phase, most of the women feel so energized. It gives you intense workouts. It helps you giving an intense workouts. And when you are in a menstrual phase that you will be having low recovery rate and you don't have mood to do some workouts, it adjusts accordingly. And the workout and the diet nutrition and the nutrition notifications are also very much helpful for the people who are in the in the different phases. Like, let's say, in an example, if you are in a menstrual phase, you want iron more because you are you will be losing more blood. So it suggests you take 10% more iron foods, for example, let's say so and so, so and so, fruits, so that you will be recovered from the pain and all. So these are some kind of stuff that I'm very proud of.

Dennise Cardona  12:59:04  

I have to just ask, what made you decide to go with that angle? Because it's, it's not something might not be intuitive to you as a male. It may it may be, but it may not be.

Manikanta Sirumalla  12:59:14  

Yeah, even I was very much like in a question mark, like in a question mark phase, where should I keep it or not? Because I don't know much about the female reproductive system in the real world. Data. We can go through the books and all, but we don't know how people relate it in the real world, right? So one thing I did was I asked some of my friends, some of my female friends, into fitness, and I given a Google form. I generated a Google form with different questions and all. I just distributed it all over my friends in the US, in the in the India, in different different people, different universities. Then I collected the collective thing, and the average baseline out of it, the what they wanted, actually, how they wanted to be a fitness app that related to their menstrual cycle. So that is a background work. All went out there for the app, for the feature, speaking

Dennise Cardona  13:00:04  

again about the technical aspects and all of that, what's the main problem that rep Track Pro solves for people in their day to day life? 

Manikanta Sirumalla  13:00:14  

Yeah, it's actually simple. Like most of the people who wanted to get in shape hit the same wall, like I do, right? Don't know what to do, how to do, where to do, and all they simply go for go to the gym and sign up, and they watch some YouTube videos, and they stand in front of the rack and barbels, and they think, like, hoof, what's now, how to do and what to do, and I think, how many sets and all that said, that sort of questions that beginner, beginner, gonna have when he's first stepped out, stepped into the gym, and that, Like that's a wall that reprac Pro breaks through, you know, like it builds you once you give yourself all the data, like your proficiency in lifting, your your weights and your diet preferences, like, let's say you are vegetarian, non vegetarian, or mix of vegan, keto, or something, all the all that sort of things. It builds, it builds you the programs, and it builds you the nutrition accordingly, with the videos, with all the things and what not, it's, it's everything like, you know, in the nutrition also, we have this thing. Let's say you cheated. You have a cheat meal a day. Let's say Chick fil A. You have some some strips, or something, that sort of thing. I've added USDA. USDA certified API, where you can search for the food. It directly gives you completely calories, macros and all this. You can add it, and you can track through the nutritional level.

Dennise Cardona  13:01:29  

That's great. I was thinking yesterday I had a big cheat day. I ate a lot of Colombian bakery, baked, baked goods. And I don't even know that I'd want to track all of that. Now. It's good. 

Manikanta Sirumalla  13:01:43  

Let's say if you're serious in a 12 week straight program, let's say so if you wanted, really wanted to change for that 12 weeks you want progress, then you have to track, right? You have to be very, very particular. Even, let's say a bodybuilders who are in competitive field, they even track the amount of water they drink, the amount of spoons of water that drink, so they will be. That particular so I wanted to deliberately add that feature in the in the app. So I love it.

Dennise Cardona  13:02:08  

That's good. So I have one last question for you, looking ahead, where do you want to take rep track, pro next, and how do you see your career evolving after you are done studying here at UMBC?

Manikanta Sirumalla  13:02:22  

Looking ahead, the next mile. The next big milestone for rep Track Pro is in few weeks, I'll be launching the pro tire where 999, a month, and 7999 and a year. And that's a big moment that I'm looking for. And I've also applied for the UMBC Launchpad accelerator program. And I think that that by Kevin's, by Kevin's and grace and all. If he selects me, and if some funding may happen, then I'll be, then I'll be working more on this one, to taking it to the real world and coming to UMBC. I do. I don't want to be in the UMBC bubble for the for the application. I wanted to go out, spreading to the different, different universities around the regional universities and all. And, yeah, I think rep Track Pro feels good in the future, and I'll be working more on this one like I think this taught me so many great things that even if I worked for these three years into instead of taking masters, I wouldn't have learned so much. So it's good, and reptrack Pro is good enough. And also pitching the Y Combinator, so we don't know by God's grace if any any great things happen, and this will be the next big thing.

Dennise Cardona  13:03:31  

So happy to hear that. Thank you so much. So so much for sharing your insights and sharing the great news about your app and how you how you created it and how you got it out there. So happy to hear that such a great story, success story from UMBC. It's wonderful to hear so thank you so much for sharing that. Thank you so much, and thank you so much to our listeners for tuning into this episode of UMBC miked up podcast. If you'd like to learn more about our offerings and some of the things we talked about today, like rep Track Pro, I will be putting that in the show notes, so check those out. Thanks so much.