Curious about people.Fluent in funnels.Building with AI.

Two companies, one question

From fintech to proptech, my work keeps coming back to the same question: where do people get stuck, and why? Here's how that played out.

1. 2024 to 2026, INDmoney

   Growth hides in small, hidden frictions

   Leading investor growth, I kept finding that drop-offs had simple, hidden causes. Our unified investing campaign brought in users for both Indian and US markets, but they couldn't all be sent down one KYC journey. So I let them choose, and KYC completion rose 1.8x and investor conversion 1.1x. Referrals needed better incentives and clearer reward tracking, so I rebuilt them. Then regulation shut that channel down, and I moved growth to credit-card bill payments instead.

   The obvious problem is rarely the real one, and constraints can point you somewhere better.

2. 2026 to present, Housing.com

   Making trust the easy path

   I own platform health for the marketplace: verification, listing quality and seller responsiveness. Only 10% of brokers verified listings, not because they didn't care, but because it meant visiting every property. I made it possible from home, with an ML audit to keep it honest. I tuned the thresholds with our ops team and talked to 20 brokers before redesigning how we measure responsiveness.

   Start small, listen to the people closest to the work, and earn the right to scale.

3. Now, on the side

   Learning to build with AI

   Setting accuracy thresholds taught me that the hard part of an AI product isn't the model. It's deciding how wrong is acceptable. Now I want to build that kind of thing myself. I use Claude Code for almost everything, from digging through data to sketching a prototype at midnight, and I'm building small AI tools to learn by doing.

   The next step is AI that takes messy, manual work off people's hands, with trust built in from day one.

How I work through a problem

The same five steps sit behind every case study below. I use them to stay honest about whether a problem is real, worth solving, and actually solved.

1. Find the real problem

   A drop-off is a symptom. I look for the cause in the funnel data and by talking to the people stuck in it.

   Seen in: KYC routing, broker interviews

2. Check it's worth solving

   Before building, I size the gap against a benchmark or a cost we already pay.

   Seen in: competitor verification benchmarks, manual calling costs

3. Make one clear bet

   A hypothesis I can prove wrong, tied to a single metric that tells us if it worked.

   Seen in: letting users choose their KYC journey

4. Test small, decide the trade-off

   One city, one cohort or one A/B test. Small tests are where I find out what a trade-off actually costs.

   Seen in: the one-city verification pilot

5. Scale, then learn

   Roll out once the numbers hold, refine with the people closest to the work, and write down what I'd change.

   Seen in: pan-India rollout with ops

What that looked like in numbers

Three funnels, before and after.

Case studies

Six problems, each walked through from the real problem to the result. Pick one to read the full story.

Housing.com · Platform health

Verification

No Site Visit Required

Only 1 in 10 brokers verified their listings. They wanted the badge. They just couldn't get to it.

12% → 18% verified listingsRead case study → Listing quality

What Makes a Listing Good?

We had no way to tell a great listing from a bad one. So we built a score, then gave sellers a reason to chase it.

+20% listing qualityRead case study → Seller responsiveness

Left on Read

Buyers said brokers never called. Brokers said buyers never picked up. Everyone was paying for it.

+20% response rateRead case study → AI voice bot

One Enquiry, Many Homes

A buyer who asks about one home is often a fit for three more. We built a voice bot to make sure someone always asked.

3K/mo new leadsRead case study →

INDmoney · Investor growth

Onboarding & KYC

Two Markets, One Door

Our ad promised Indian and US stocks. Our onboarding could only send people down one road.

1.8x KYC completionRead case study → Referrals

When the Rules Changed Overnight

We rebuilt a referral program until it worked. Then regulation shut it down, and we had to build it again.

6.2x sign-upsRead case study →

Before product

Where I first learned that systems are really about people.

Off the clock

Same curiosity, different settings.