FIN-AI: Enterprise Financial Intelligence & Algorithmic Allocation Platform

FIN-AI is an end-to-end full-stack financial suite designed to synthesize portfolio management, debt restructuring, market technicals, and generative reasoning. Built with a decoupled FastAPI backend and React (Vite) frontend, it integrates cryptographic tenant isolation, multimodal receipt ingestion, and algorithmic graph simplification.

---

Key Modules & Core Capabilities

• Financial Health Index & Predictive Runway:

  • Real-time 0–100 composite diagnostic derived from three financial pillars: Savings Surplus Rate, Reserve Runway (months of expenses covered), and Operating Burn Ratio.
  • 30-day forward-looking predictive cash flow horizon with cyclic variance envelopes ($\pm 12%$).
  • Automated fiscal audit powered by Google Gemini generative models.

• Algorithmic Capital Optimizer (Debt Avalanche):

  • Implements the Greedy Debt Avalanche strategy ($O(N \log N)$ complexity).
  • Automatically calculates mandatory baseline obligations, extracts surplus liquidity, and routes capital to highest-APR liabilities to minimize interest accrual over time.

• Social Debt Simplification (Graph Minimization):

  • Models peer obligations as a directed graph $G = (V, E)$.
  • Resolves multi-party circular liabilities into minimum-cash-flow transactions using net vertex balance resolution ($O(V + E)$), eliminating circular transaction churn.

• Multimodal Document Scanner (Computer Vision):

  • Ingests physical paper receipts and digital invoices using Gemini Vision.
  • Automatically extracts vendor entity, transaction date, expense category, and total committed outflow directly into the user's ledger.

• Market Terminal & Equity Technicals:

  • Live quote synchronization via Yahoo Finance API (yfinance).
  • Custom vector candlestick price charts with hover telemetry and Simple Moving Average (SMA) overlays.
  • GenAI trade execution analysis providing entry zones, resistance thresholds, stop-loss channels, and sector peer benchmarking.

• Multilingual Audio Financial Advisor:

  • Real-time Speech-to-Text (STT) and native Web Speech Synthesis (TTS).
  • Full prompt fluency across Indian scripts (Hindi, Bengali, English) with customizable response profiles (Balanced, Concise, Analytical, Action Steps).

• Executive Dossier Export:

  • One-click client-side print-to-PDF engine compiling institutional summaries, health metrics, and settlement paths into a clean vector report.

---

Security & Architecture

• Cryptographic Hashing: Salted PBKDF2-HMAC-SHA256 with 100,000 iterations for zero-exposure credential storage.
• Token Authentication: Cryptographically signed HS256 JSON Web Tokens (JWT) with 7-day session lifespans.
• Tenant Isolation: Strict relational isolation via foreign key bindings (user_id) across SQLite tables (transactions, liabilities, peer_debts).
• Brute-Force Mitigation: Sliding-window in-memory rate limiting restricting authentication endpoints to 5 failed attempts per 60 seconds per IP address.

---

Tech Stack

Domain	Technology
Frontend	React 18, Vite, Recharts, Lucide-style Inline SVG Icons
Backend	FastAPI, Uvicorn, Pydantic, Python 3.10+
Database	SQLite, SQLAlchemy ORM
AI / Multimodal	Google Gemini API (google-generativeai), Pillow (PIL)
Market Data	Yahoo Finance (yfinance)

---

Project Structure

Financial_Assistant_PBL_Project/
├── client/                     # React + Vite Frontend
│   ├── src/
│   │   ├── App.jsx             # Unified interface orchestrator & modules
│   │   ├── App.css             # Dark-mode terminal design system & print styles
│   │   └── main.jsx
│   ├── package.json
│   └── vite.config.js
├── server/                     # FastAPI Backend
│   ├── .env                    # Secrets (JWT_SECRET, GOOGLE_API_KEY)
│   ├── finance.db              # SQLite persistence layer (auto-generated)
│   └── main.py                 # Secured REST endpoints & database models
├── archive/                    # Archived early prototype iterations
├── .gitignore                  # Production Git exclusions
└── README.md                   # System documentation