Work

Three teams, three very different product surfaces. Each role added a new layer to how I think about data, decisions, and the products that sit between them.

GyanDhan

Bringing the same decomposition-and-experimentation rigor I built at Loop and Apna Mart to education-lending decisions. The focus is on making product analytics a lever the team reaches for, not a report they skim once a week.

It's early, but the appetite is there — which is most of what matters.

Loop

Built decomposition frameworks to track reimbursement-claims performance. The aggregate query rate looked steady; the segments did not. Breaking the volume down by TPA and disease category surfaced a small set of concentrations, and acting on those — not the average — cut the query rate from 40% to 32%.

Same method with turnaround time: we targeted P90 on slow claims directly rather than the mean. Surfacing disease-wise document patterns across top TPAs cut P90 TAT by 21%. The average barely moved. That was the point.

Apna Mart

Owned end-to-end product experiments — design, measurement, monitoring, and rollout. Performed deep root-cause analysis on business-critical problems, with a bias toward shipping fixes that changed what the team did the following week.

The habit I took from Apna Mart: if the metric isn't measurable with the instrumentation you already have, that's a finding, not a blocker.