FlowWatch AI : Live

FlowWatch AI is a full-stack banking intelligence platform for real-time fund tracking and fraud detection. The project combines a FastAPI backend, an AI/ML risk scoring engine, graph-based flow tracing, STR reporting workflows, and a responsive React frontend tailored for banking investigation teams.

Overview

• monitors suspicious transactions in near real time
• scores fraud risk using machine learning and graph signals
• traces fund movement across linked bank accounts
• generates alerts with AI-assisted explanations
• supports suspicious transaction report workflows
• provides dashboards, analytics, and investigator tooling

Repository Structure

FlowWatch AI/
  README.md
  PROJECT_DOCUMENTATION.md
  PROJECT_REPORT.md
  FlowWatch_AI_PSBs_Hackathon.pptx
  flowwatch-backend/
  flowwatch-frontend/

Key Modules

Backend

Located in flowwatch-backend

• FastAPI REST API
• PostgreSQL, Redis, MongoDB integration
• Celery workers for async processing
• fraud scoring pipeline with Isolation Forest, XGBoost, NetworkX, and SHAP
• JWT authentication
• seeded demo data and model training flow

Frontend

Located in flowwatch-frontend

• React + Vite operator console
• responsive dashboard and investigation screens
• alerts panel and real-time notifications
• fund-flow graph visualization
• STR, profile, and settings pages
• Vercel-ready frontend deployment setup

Core Features

• Real-time transaction monitoring with fraud risk scoring
• Dashboard for transaction volume, alert counts, and amount at risk
• Live alerts with severity tracking and investigation workflows
• Interactive fund-flow tracing using graph visualization
• STR generation and report management
• Analytics for fraud type trends, accuracy, bank risk, and heatmaps

Quick Start

Backend

cd "/Users/anurag/Desktop/FlowWatch AI/flowwatch-backend"
cp .env.example .env
docker compose up -d postgres redis mongodb celery_worker celery_beat
docker rm -f flowwatch-backend-8001 2>/dev/null || true
docker run -d \
  --name flowwatch-backend-8001 \
  --network flowwatch-backend_default \
  -p 8001:8000 \
  --env-file .env \
  -v "$PWD:/app" \
  flowwatch-backend-backend \
  uvicorn app.main:app --host 0.0.0.0 --port 8000

Frontend

cd "/Users/anurag/Desktop/FlowWatch AI/flowwatch-frontend"
npm install
npm run dev -- --host 0.0.0.0 --port 5174

Access

• Frontend: http://localhost:5174
• Backend Docs: http://localhost:8001/docs

Demo Login

• Username: anurag@rbi.in
• Password: Anurag@1432

Deployment Summary

Frontend

• recommended target: Vercel
• vercel.json already included
• set:
  • VITE_API_BASE_URL
  • VITE_WS_BASE_URL

Backend

• deploy separately on Docker, VM, or container infrastructure
• requires persistent:
  • PostgreSQL
  • Redis
  • MongoDB
  • Celery worker
  • Celery beat

Documentation

• Full technical notes: PROJECT_DOCUMENTATION.md
• Project summary/report: PROJECT_REPORT.md
• Frontend-specific guide: flowwatch-frontend/README.md

Team Details

Name	Role	Responsibility
Anurag	Project Owner / Maintainer	Full-stack development, integration, deployment readiness
Yash	Member	Data Engineer
Prakhar	Member	System Design
Reeshav	Member	MLOps

• Repository: Anurag-M1/FlowWatch-AI