Tushar’s story

Fixing Paytail's Loan Interest Calculation End to End

Identified a loan compounding error, built the math and code solution, and shipped it into the product with cross-functional coordination

Tushar Vigh

Junior Analyst at JPMorganChase

JJPMorganChase
IIFSA Network
IIFSA BITS Pilani
DDepartment of Reception and Accommodation
AAPOGEE, BITS Pilani
4+ years of experience

From their time as

P

Product intern

Paytail β€’ 2022 - 2022

Overview

Tushar joined Paytail, a fintech startup, as a product intern. One of the problems he was handed was a client complaint: the interest calculation for the first loan period was coming out differently from subsequent monthly cycles, because the first period ran from the loan origination date to the billing cycle date rather than a full month.

The Story

Tushar joined Paytail, a fintech startup, as a product intern. One of the problems he was handed was a client complaint: the interest calculation for the first loan period was coming out differently from subsequent monthly cycles, because the first period ran from the loan origination date to the billing cycle date rather than a full month.

He started with pen and paper. He worked through the math manually first, mapping out how the interest should compound across the irregular first period and the regular monthly cycles that followed. Once he had the logic right on paper, he translated it into code.

With the code written, he worked with the backend team to integrate it into the platform. In parallel, he worked with the product team on how to surface the calculation clearly in the app interface, and with the client-facing team on how to explain it to users. He framed the fix as a USP: showing the interest breakdown transparently at the top of the loan summary.

The fix shipped into the product. Tushar received a letter of recommendation from the founder for the work.

Separately, he also identified a bottleneck in the loan approval process: sales reps were sending customer data in unstructured formats, requiring manual cleaning before it could enter the data system. He built an automated pipeline to extract structured data and feed it directly into the backend. The process went from a full-day turnaround to roughly one hour.