AI-Bank-Statement-Document-Automation-By-LLM-And-Personal-Finanical-Analysis-Prediction

Why we included this project

Bank statements arrive as messy PDFs, and this project is a full working pipeline for turning them into structured data you can actually query. YOLO layout detection and OCR find the tables, then an LLM extracts the transactions, sorts them into income and expense categories, tracks spending trends, and answers natural-language questions about the results. One design choice worth noting: it redacts personally identifiable information before anything gets embedded into the vector store, which matters if you're handling real customer documents. The repo also includes a complete FastAPI + PostgreSQL + Celery + React service with REST auth and async processing, so it runs as a real application rather than just a notebook. If you're exploring agent-based document workflows, the parallel CrewAI, Deep Agents, and Hermes harnesses all tackle the same bank-statement task, and there's practical guidance for running local models through LM Studio or Ollama.