The last two years of building a company have taken me in directions I never would have imagined.
We've explored different ideas, gone through several pivots, learned a tremendous amount, and more recently, navigated a transition. I'm grateful to Spencer, with whom I started this journey, and to everyone who's been part of it. These two years have taught me far more than I could have anticipated.
We're building Shifu as a personal health assistant that understands your health context over time, helps with everyday accountability, and connects you with care professionals when needed.
And as I've spent some time reflecting on the journey so far, I've realized just how consistently I've found myself drawn to this problem over the past decade.
It started with a seminar in 2017
I remember attending a seminar in 2017, barely a month into my freshman year at IIT Indore.
I was just exploring what was happening around campus. This particular seminar was about the state of computer vision in healthcare, especially applications in medical imaging and radiology, and where the technology might go in the coming years.
Some of those predictions have since become reality.
Something about the potential impact fascinated me. The idea that computers could learn to interpret medical images, recognize patterns, and help us understand the human body in ways that weren't previously possible was incredibly exciting.
I ended up orienting much of my undergraduate journey around the intersection of computer vision and healthcare.
Along the way, I worked on a diabetic retinopathy project that received recognition from BETiC at IIT Bombay. I also spent several months building an assistive system for visually impaired people with my friends Suraj Polamaina and Shavez Malick.
That second project was particularly memorable. We spent around 6–8 months working on it, putting together prototypes, making our first pitch decks, and even trying to raise money for an idea. The project eventually received national recognition from Samsung and Siemens Healthcare.
Looking back, it was probably my first taste of entrepreneurship, even though I didn't think of it that way at the time.
Back then, I imagined becoming a researcher or professor working at the intersection of AI and healthcare. I followed researchers like Jitendra Malik at UC Berkeley closely and dreamed of doing research in that environment.
What kept me going was the possibility of impact. We'd see what the technology could do, imagine what it could mean for people, and that alone would fill us with energy.
When technology met real healthcare
A few years later, I was supposed to head to France to work at a cancer research institute on computer vision applications.
Then COVID happened, and those plans fell through.
Trying to make the most of the situation, I got an opportunity to work with Dozee on remote patient monitoring during the pandemic.
We were building technology that could help convert ordinary hospital beds into step-down ICU beds, enabling continuous, contactless patient monitoring.
I remember diving deep into the underlying technology, particularly ballistocardiography, and being amazed by how much information we could derive about someone's health from subtle signals produced by the human body.
One thing I found particularly fascinating was that ballistocardiography wasn't new. It had existed for decades, but advances in sensors, computation, and AI were creating new possibilities for interpreting those signals.
Sometimes, new technology gives us an entirely new lens through which to look at something we've known about for a long time.
That realization stayed with me.
Working at Dozee was also an opportunity to watch Gaurav and Mudit build a company from the ground up. Seeing the energy, uncertainty, and impact of that zero-to-one journey made me want to experience it myself someday.
When I later moved to Georgia Tech for my master's, one of my goals was to work with the Wellness Technology Lab.
I finally got that opportunity in my second year, working on women's health, eventually leading the development team, collaborating with Emory, and working alongside other health technology startups.
This was also my first real exposure to conducting study groups and small trials, and understanding how healthcare research and delivery work in practice, particularly in the US.
One of my biggest takeaways was how different healthcare needs can be across populations.
Women's health, for example, presents very different needs and challenges from men's health. The kind of support that works for one group, or even one individual, might not be appropriate for another.
It made me appreciate that building useful healthcare technology isn't just about solving a technically interesting problem. It's about understanding who you're building for, the context in which they live, and how the technology fits into their lives.
Two years of building, jumping, and learning
I've always wanted to be part of building something from zero to one.
A big part of my motivation for starting a company was probably naïve optimism. I'd seen what building a company looked like from the outside, and I wanted to experience it firsthand.
I wanted to work on something that gave me so much energy that work wouldn't really feel like work.
The last two years have taught me a lot about what that actually takes.
We've explored several directions, made mistakes, and learned some lessons the hard way.
One of the biggest lessons has been understanding the difference between a pivot and a jump.
A pivot builds on what you've learned. A jump takes you somewhere entirely different, often without carrying enough of those learnings forward.
Looking back, I think we jumped more often than we pivoted.
There's a place for intuition in building a company. Sometimes you have to trust your gut, especially when you're working on something that doesn't exist yet.
But I've also learned that enthusiasm and conviction alone aren't enough to make strategic decisions. You need evidence, conversations with users, and a willingness to confront what the data is telling you, even when it doesn't align with what you'd hoped.
I don't think this means I'll never pivot again. Quite the opposite.
What feels different today is that I'm increasingly committed to the problem space, rather than any particular solution.
I want to spend years understanding how care can be delivered better. The products and approaches we take might evolve, and I'm comfortable with that.
Why health has always been personal
There's another side to this story that I've come to appreciate more in retrospect.
Fitness and wellness have been a big part of my life for years.
Through some of the most difficult periods, having a consistent routine has helped me stay grounded.
There's something about getting into a Muay Thai sparring session where, for that period of time, nothing else really matters. You're completely focused on what's immediately in front of you. And when you come out, everything else somehow feels a little simpler and more manageable.
Even today, making it to the gym or a Muay Thai session is one of the most important parts of my day.
I've experienced firsthand how much taking care of your health can influence everything else in your life.
And yet, despite caring deeply about my health, I went nearly two years without seeing a doctor after moving to the US.
Not because I was deliberately avoiding it. I generally felt healthy, but finding the right professional, navigating the system, figuring out when to seek guidance, and making the time for it all felt like more friction than it should have been.
Looking back, I probably could have benefited from professional guidance during those years.
I find that contradiction interesting.
If someone who cares deeply about health still struggles to engage with the healthcare system, surely there's room to make it easier.
And not every interaction with a healthcare professional should require scheduling an appointment, traveling somewhere, or sitting through a call.
Sometimes you just need someone who understands your situation to point you in the right direction.
What we're learning with Shifu Health
Over the past several months, we've been building Shifu Health around a relatively simple belief: care should be easier to access, more continuous, and more efficient to deliver.
We're still early, with a small group of people who have been using Shifu consistently for months.
Some of the things we've seen have genuinely surprised me.
We've had users log more than 500 meals over a couple of months while working toward their weight-loss goals. I hadn't anticipated that someone would engage with the product so rigorously.
It's been fascinating to see how much consistency some people are willing to bring to their health when they have a system that fits into their daily lives.
One experience that's been particularly meaningful is seeing a close friend use Shifu as part of his journey back to working out after shoulder issues.
He'd faced challenges with rehabilitation before, and it's been encouraging to see him finding value in having an ongoing accountability partner as he works toward returning to exercise, with professional guidance accessible when needed.
There's something special about seeing technology you've worked on become genuinely useful in the life of someone you care about.
These experiences have reinforced a belief I've developed over the years:
Care shouldn't have to start from scratch every time you need help.
Our health has context. Our routines, activity, nutrition, recovery, goals, medical history, and all the small changes happening over time.
Today, we have more ways to capture this information than ever before. Wearables, smartphones, and connected devices give us an increasingly detailed picture of our everyday lives.
But making that information useful, both for the person seeking care and for the professional delivering it, remains remarkably difficult.
That's where I think AI can make a meaningful difference.
Not just by answering health questions, but by helping maintain context over time, supporting everyday accountability, and making it much easier for professionals to understand someone's situation.
One of the things that has been encouraging is how strongly the care professionals we work with believe in the potential of asynchronous care.
Not every interaction needs to be a meeting or a phone call.
A significant amount of guidance and follow-up could happen asynchronously, provided the professional has sufficient context to make an informed judgment.
And that's a big part of what we're exploring with Shifu.
What if a care professional could get caught up on months of someone's health context in minutes, and provide useful guidance without requiring an hour-long appointment?
I believe that in a world of increasingly capable personal AI agents, healthcare will remain an area where human judgment matters enormously.
AI may become incredibly knowledgeable, perhaps even exceeding human experts in many specific tasks. But delivering care involves much more than knowledge alone. It involves understanding individual circumstances, weighing uncertainty, making decisions, and taking responsibility for them.
I see AI making care professionals substantially more effective, not eliminating the need for them.
There are still many unanswered questions.
Building lasting habits is hard. We're learning how to make accountability genuinely useful without becoming intrusive, how to help people stay engaged over time, and where human intervention makes the biggest difference.
For now, we're particularly focused on people who already recognize gaps in how they manage their health and are actively trying to improve it.
Our hope is that Shifu can become a longitudinal health assistant and accountability partner, with professionals available when their expertise is needed.
There's a lot left to figure out.
The next chapter
When I look back at that seminar in 2017, my fascination was largely with what technology could understand about the human body.
At Dozee, I learned how much information we can extract from the human body, and how advances in technology could make old ideas useful in new ways.
At Georgia Tech, I started appreciating the complexity of healthcare delivery and how different people's needs can be.
And through building a company, I've learned that creating something people genuinely value requires much more than an exciting technology or a compelling idea.
With Shifu, we're exploring how to turn that information and longitudinal health context into useful, ongoing care.
The last several months of building Shifu, learning from users, and working with care professionals have only strengthened a conviction that's been forming for nearly a decade.
I don't know exactly what Shifu will look like a few years from now. But I've become increasingly committed to this problem space, and I'm excited to keep learning what actually works.
The next chapter is about bringing Shifu to more people, learning from them, and continuing to explore how care can be easier to access and more efficient to deliver.
Looking back, the dots connect.
Excited for what comes next.