Amanda Hsu

An agent that lives where we already reach students: their phones.

Students reply directly to billing reminders with questions or notes. Agent conversations save schools time and surface signals that feed into risk scoring, ultimately helping staff answer: who needs my attention today?

Most recently, I’ve been building an eval framework for the agent — starting with hands-on error analysis to define what an excellent response looks like, then automating it into a continuous improvement system.

A campaign management system that allows school admins to preview, manage, and test the billing reminders students receive through Meadow. This was one of the most highly requested pages and later became the foundation of our campaign upsell strategy.

This was also the first project where I coded the entire frontend in React, a workflow that's since become how we ship every project!

A filtering interaction that detects student filters and exclusions in real time as school admins type out (or dictate!) a query.

Core to the school admin's role is pulling a precise, multi-attribute student segment to analyze or act on. This exploration was born from the observation that the process involves far too many clicks (or is entirely impossible in some cases).

Launched our first subscription product, including a customized cash back program that rewarded users with 5% back at 5 places they pick.

Increasing Savings Jar adoption among new users through contextual feature education and thoughtful touches of motion design.

Redesigned our leaky onboarding funnel and implemented a dynamic, risk-based identity verification flow, increasing application approvals by 62%. A fascinating deep dive into KYC and AML regulation.