Case Study · 2024

Finance Tracker

Personal budgeting with delightful data viz

Next.jsPostgreSQLPrismatRPCD3.js
Finance Tracker cover

Overview

A privacy-first personal finance app that helps people categorize spending, set budgets, and visualize money flow over time — without selling their data or pushing credit products.

The Challenge

Importing transactions from dozens of bank formats and keeping categorization accurate without sending raw data to third parties.

The Solution

I built a local-first ingestion pipeline with deterministic rules plus an opt-in on-device ML model for fuzzy categorization. The Next.js app uses Server Components for fast loads and tRPC for typed mutations.

Key Features

  • CSV/OFX import with smart bank format detection
  • Customizable budgets with rollover and alerts
  • Interactive sankey and time-series charts
  • Recurring transaction detection
  • Encrypted backups to the user's own cloud

Results

  • 4.8★ average user rating in private beta
  • Median import flow under 30 seconds
  • Zero PII leaves the device unencrypted