FinAI Sentinel: AI-Powered Personal Finance System

This paper presents FinAI, an AI-powered personal finance management platform that uses cutting-edge machine learning and deep learning methods to automate tasks like monitoring finances, analyzing behavior, making predictions, and optimizing budgets. Many people still have trouble managing their money because they have to use different tools, keep track of their spending by hand, and don't have access to advanced decision support. Most of the finance apps that are out now use static budgeting methods and rule-based systems, which makes it hard for them to adapt to how users change. The proposed system integrates several specialized models, including FinBERT for transaction classification, Graph Neural Networks (GNN) for spending behavior analysis, Temporal Fusion Transformers (TFT) for multi-horizon expense forecasting, Temporal Graph Networks (TGN) for anomaly detection, and a Deep QNetwork (DQN) reinforcement learning agent for dynamic budget optimization. The system also includes optical character recognition (OCR) and voice-based spending entry, which lets you enter data in more than one way. SHAP-based explainable AI is used to explain financial health scores and model projections in order to make things clearer and give customers more trust. A PostgreSQL database, a Next.js frontend, a FastAPI backend, and Celery-based asynchronous AI pipelines make up the full-stack architecture of the platform. The system works well when tested with a synthetic dataset of 50,000 transactions. The F1-score for categorizing transactions is 97.3%, the AUC for finding anomalies is 94.1%, and the MAPE for estimating costs is 8.4%. Reinforcement learning optimization also helps save 23% more money. The results show that FinAI is a smart, scalable, and easy-to-understand solution for future personal finance management systems.