Final YearAdvancedAI/ML 16-20 weeks
AI-Powered Personal Finance Manager
A comprehensive personal finance platform that connects to bank accounts via Plaid, auto-categorizes transactions using ML, predicts future spending, and gives AI-driven savings recommendations.
Next.jsPlaid APIPython (FastAPI)scikit-learnPostgreSQLRedisOpenAI
Find GitHub repos & source code →How to build it — step by step
- 1Bank Integration: Plaid Link for bank connection; fetch transactions, balance, and account metadata via Plaid API
- 2ML Classification: Train a multi-label classifier (XGBoost) on transaction descriptions to auto-categorize spending
- 3Prediction: Time-series forecasting (Prophet/ARIMA) to predict next month's spending by category
- 4AI Advisor: GPT-4o with function calling to answer "How can I save more?" with context from actual financial data
Key features to implement
- ✓Real bank account connection (Plaid sandbox)
- ✓Automatic transaction categorization (95%+ accuracy)
- ✓Monthly spending forecasts by category
- ✓Personalized AI financial advice
- ✓Net worth tracking over time
💡 Unique twist to stand out
Build a "Financial Twin" — an AI model trained on the user's specific spending patterns that simulates financial scenarios: "What happens to my savings if I reduce dining out by $200/month?"
🎓 What you'll learn
FinTech API integration, ML classification pipeline, time-series forecasting, and AI-augmented application design.