Hospitals don't need more software. They need software that fits how they actually work.
I spent months inside hospitals before writing a single line of a spec. Not observing from the outside, but actually inside on the floor learning how care runs versus how it reads on paper. That gap became the product.
I owned the product from discovery to adoption, building operations and preventive-care software that hospital teams used daily. After three years of robust growth, the company was acquired by Zocdoc in 2020.
The skills that actually shape your career aren't taught anywhere.
We built a community-first learning platform to fix that. It started as a WhatsApp group and quickly grew to 10,000 users in six months. We launched a mobile learning app that hit 100k+ watch hours and scaled the core community to over 3,500 active members.
After talking to users, we pivoted to live cohorts. We maintained a 40% completion rate (the industry average is 5-10%) by adding gamification that lifted session times by 50% and doubled retention. We ran this tight ship with just a 3-person team, producing ₹15L in revenue while testing a B2B model alongside 2 companies.
We shut it down in 2023 when the unit economics didn't work at scale.
Products and experiments that didn't survive contact with reality. Documented here because the failures are often more instructive than the wins.
The Hypothesis: Developers spend too much time manually syncing environment variables across microservices. A CLI tool could handle this securely via end-to-end encryption.
The Execution: Built a rust-based CLI and a basic web dashboard in 3 weeks. Got it working perfectly for my own local environment.
The Reality: I tried to onboard three engineer friends. Two of them couldn't install the CLI due to weird pathing issues on Windows, and the third said, "I just use Doppler for this." I had built a feature, not a product. I killed it before ever launching.
The Hypothesis: Junior designers want 15-minute, highly tactical feedback sessions on specific Figma files, not months-long mentorships.
The Execution: Spun up a marketplace. Manually recruited 10 senior designers. We hit 100 sessions booked in the first month. Incredible early traction.
The Reality: Junior designers were willing to pay $15 for feedback. Senior designers wouldn't get out of bed for less than $100/hr. The marketplace suffered from a total mismatch in pricing expectations. I shut it down rather than trying to force the unit economics to work.