Shastra 2026 — Transaction Intelligence

AI-powered EDA, anomaly detection, and insights for bank transactions.

What it does

• Upload a CSV of bank transactions and get instant EDA, anomaly detection, and AI insights.
• Consolidated anomaly report with high-risk columns, type validation errors, and ML outliers.
• Frontend dashboard (Vite + React) consuming the FastAPI backend.

Quick start

# Backend (FastAPI)
cd backend
pip install -e .
uvicorn api:app --reload --port 8000

# Frontend (Vite React)
cd ../frontend
npm install
npm run dev

Try the API quickly

curl -X POST "http://localhost:8000/analyze-transactions?num_samples=5" \
  -H "Content-Type: multipart/form-data" \
  -F "file=@path/to/transactions.csv"

Key endpoints

• GET /health — service heartbeat
• POST /analyze-transactions?num_samples=5 — upload CSV and receive full analysis

Frontend entry points

• frontend/src/App.tsx — main dashboard shell
• frontend/src/components/upload/AnalysisStepper.tsx — upload + run analysis
• frontend/src/components/anomaly/AnomalySummary.tsx — anomaly highlights
• frontend/src/lib/api.ts — API client

Data samples

• Sample CSVs: data/*.csv (multi-bank variants included)