TaxWow

🧾 About the Project – TaxWow

💡 Inspiration

Every year, tax season brings confusion, especially for freelancers, students, and small business owners who juggle multiple sources of income and expenses. We realized there wasn’t a simple, intelligent tool that could help non-accountants file taxes confidently. That’s what inspired us to build TaxWow—an AI-powered assistant that makes taxes understandable, accurate, and less stressful.

🧠 What We Learned

Through this project, we deepened our understanding of:

Retrieval-Augmented Generation (RAG) and how to apply it to real-world use cases
The U.S. tax system and IRS deduction categories
AI-powered document parsing and vector databases
Designing multi-step pipelines with modular architecture

We also learned the importance of building user-friendly interfaces that translate complex logic into plain, actionable insights.

🛠️ How We Built It

Frontend: React + TailwindCSS for a clean, responsive UI
Backend: FastAPI for API endpoints and data processing
AI Integration: OpenAI + RAG pipeline to classify expenses and generate personalized advice
PDF Parsing: Used AI-powered document processing to extract structured data from bank statements and receipts
Vector Store: Embedded IRS tax rules and deduction examples into a retrievable knowledge base

⚠️ Challenges We Faced

Parsing PDFs with inconsistent formats was tricky, especially for scanned images—so we added a confirmation step to validate OCR results
Tuning RAG prompts to ensure tax advice was accurate, safe, and grounded in IRS rules
Integrating multiple tools (OpenAI, vector stores, OCR libraries) smoothly within a limited timeframe

Despite the challenges, building TaxWow taught us how powerful AI can be when paired with real-world context—and how it can truly empower users.