Customers wouldn't act on a payment request pushed to their phone by a collections agent, even a legitimate one. The fix wasn't better copy on the notification. I changed the format: agents generate a QR code, customers scan it on their own terms, using whatever UPI app they already trust. Adoption took off almost immediately.
Field agents were visiting customers with day-old data. They had no way to know if someone had paid another loan that morning, or abandoned a payment an hour before they knocked. I designed a config-driven data layer that pushes near-real-time signals into the app without a code release. What used to take a sprint now takes a day.
The consumer app had 18+ notification types, 14+ nudges, and 6+ pop-ups pushing delinquent customers toward payment. Self-serve rates were still low. I mapped the full lifecycle of 36L customers and found the problem: every customer saw the same experience regardless of where they were in their loan journey. One landing page, built around that gap, changed that.
Every team has a channel strategy. Almost none have a journey strategy. That gap is where customers fall out.
When adoption is low, the instinct is to fix the flow. Usually the real problem is trust or format, not friction.
Nobody gets excited about a platform. The trick is talking about what it makes possible: a two-day job instead of a two-month one.
Before product, I was an engineer. That instinct for where systems actually break still shapes how I build today.
At Bajaj Finance, I spend most days thinking about how millions of people, each at a different point in their loan journey, repay, get served, and stay on track: what they use, and how they pay. Off the clock: reading about new tech, bad puns, and the quiet conviction that infrastructure is the most romantic thing you can design.