Back to projects
Fullstack
Sep 2024 - Dec 2024

AI Finance Dashboard

An intelligent dashboard for personal finance management, featuring AI-powered spending insights and predictions.

Overview

A full-stack application that helps users track their expenses, categorize transactions automatically using machine learning, and predict future spending habits to help them save better.

Challenges

  • 1

    Building a responsive and intuitive data visualization interface handling large datasets in the browser.

  • 2

    Integrating a Python machine learning service with a Node/Next.js frontend efficiently.

  • 3

    Ensuring high performance for complex data aggregations on the database side.

Architecture & Strategy

[ Architecture Diagram Placeholder ]

Next.js handles the frontend and BFF (Backend-for-Frontend) API routes. A separate Python FastAPI service runs the ML models for categorization and forecasting, communicating via REST.

Results

Achieved >95% accuracy in automatic transaction categorization.

Processed over 100k transactions per user without UI lag.

Received positive feedback from initial beta testers regarding ease of use.

Lessons Learned

  • Separating heavy computational tasks from the main API improves user experience.
  • Client-side caching is crucial for interactive data visualization.
  • Prioritizing UX in data-heavy apps makes complex information accessible.

Tech Stack

Next.js
React
Tailwind CSS
Python
FastAPI
PostgreSQL