Home
All work
About
How this site is built
The VAT model
Architecture chapter
Contact
KhatamanduIRD compliant accounting, invoicing and inventory software used by Nepali businesses.
Khatamandu AdminThe internal panel for administering the Khatamandu platform.
ArthaAn expense tracker that forecasts future spending.
Copy email addressprajwalisharan@gmail.com
Write an email
Pause motion
Reload with the WebGL 2 renderer
Reload with the WebGPU renderer
Open GitHub
Open LinkedIn

Arrow keys to move, Enter to choose, Esc to close

Artha

An expense tracker that forecasts future spending.

Role
Design and full stack development
When
Personal project
Stack
React, TypeScript, FastAPI, PostgreSQL, scikit-learn

What I did

  • Records expenses by category: food, transport, rent, shopping and bills.
  • Forecasts future spending with a linear regression model.
  • Serves the model from a FastAPI backend with PostgreSQL storage.

What it does

Artha records expenses by category (food, transport, rent, shopping and bills) and forecasts future spending from that history.

How the forecast works

The forecast is a linear regression model built with scikit-learn and served by a FastAPI backend. It is a deliberately simple model: easy to explain, quick to retrain and honest about what it can predict.

How Artha forecasts spending

React and TypeScript in front, a FastAPI service with a scikit-learn model behind it.