I’m a precision preventative medicine physician focused on South Asian health. I’m also a hospitalist. And this year, I became something I never expected: a builder.
I built 161 apps on Replit and published 47 of them. That apparently puts me in the top 0.5% of users on the platform.
I’m not a software engineer. I’m a doctor who got tired of waiting.
The Problem No One Wants to Talk About
Here’s what I kept running into: the tools we use in medicine, the calculators, the risk scores, the reference ranges, the guidelines, weren’t built for people who look like me. Or my patients.
South Asians have a fundamentally different metabolic profile. This isn’t a minor footnote. It changes everything.
BMI cutoffs don’t apply. A South Asian at BMI 23 can carry the same metabolic risk as a white individual at BMI 30. The WHO acknowledged this years ago. Most tools still haven’t caught up.
Visceral adiposity behaves differently. We accumulate fat around organs at lower body weights, often invisible on the outside but deeply dangerous on the inside.
Beta cell function follows a different trajectory. South Asians tend to have lower beta cell reserve and earlier dysfunction. By the time fasting glucose looks “abnormal” by standard cutoffs, significant damage is already done.
The biomarkers that matter most aren’t the ones being tracked. Adiponectin, a key marker of metabolic health, is often lower in South Asians, yet almost never measured in standard care.
This isn’t speculation. The research exists. But it takes years, sometimes decades, for evidence to filter into guidelines, then into EMRs, then into the tools physicians actually use. Meanwhile, people are developing diabetes, heart disease, and fatty liver at rates far higher than their neighbors, often while being told their labs look “fine.”
I watched this happen to family members. I felt it in my own health journey. And I saw it every day in my patients.
The Turning Point
In mid-2024, Replit launched their AI Agent. I’d been frustrated with the available tools for tracking my own metabolic health, so I decided to see if I could build something better.
I built a health dashboard with wearable integration. It worked.
So I kept going.