Context
When a subscriber roams onto a partner's network, two operators end up holding two versions of the same usage. Each month those versions have to agree before money moves. NeuString is Yaana Technologies' platform for that wholesale roaming business, and its users are the operators' own settlement and finance teams.
My role
I own the frontend of three modules end to end: Reconciliation, Billing & Charging Evaluation (BCE) and the Amazon QuickSight (AQS) analytics module. That covers new features, enhancements, fixes and delivery. I review code with teams in India, Ukraine and Poland and mentor junior developers.
The problem
Settlement data is dense, numeric and unforgiving. A screen that rounds the wrong way, hides a filter state, or makes two records look alike costs real money and real time to untangle. The interfaces have to make large tables legible and discrepancies impossible to miss.
What I built
Reconciliation views that compare records and surface what doesn't match. BCE screens for evaluating billing and charging. The AQS module that brings QuickSight analytics into the product. Each one is built for people who read tables all day and need to trust what they see.
Constraints
Client work under NDA: no screenshots or real data appear here. The interactive table on this page is an illustrative recreation running on synthetic data.