← Back to work

Side project · Personal finance

Stacked – Personal Finance & Budgeting App

Side project · Product owner and builder · ML-driven spend insights

What I did: Designed and built a budgeting app that categorises transactions automatically and shows where the money goes.

Result: A working app used at home, with a public demo you can try.

Summary

  • Personal finance tool designed around automation over manual budgeting.
  • ML-assisted transaction categorisation for cleaner insights.
  • System design built for open banking integrations.

Context & problem

Many budgeting tools assume users will manually maintain categories, envelopes and tags. Stacked flips this: it prioritises automation, low effort, and clear month-end insights.

Constraints

  • Built to integrate cleanly with APIs like TrueLayer.
  • Requires explainability for category/nudge decisions.
  • Mobile-first design for daily use and weekly summaries.

Approach

User behaviour insights

Aimed at users who want clarity with minimal maintenance. No spreadsheets, no heavy UI.

System shaping

  • API ingestion → cleansing → categorisation engine → insights layer → UI.
  • Custom category rules + ML suggestions with confidence scores.
  • “One-screen” summary with drilldown for anomalies.

Outcomes

  • Clear product story and user value proposition.
  • Scalable data model for real bank feeds.
  • Foundation for shared budgeting and couple-mode.

Where this applies

Built for personal use, but the building blocks are common:

  • Categorisation with confidence scores: expense management, accounting software, product tagging, support ticket routing
  • Insight-first, low-maintenance UX: any product where users won't keep data up to date by hand
  • Bank feed ingestion: lending and affordability checks, accounting integrations

Next steps

Next iterations: deeper ML clusters, shared spending flows, and explainable insights for end users.