Real-Time Fraud Detection on an HPC Cluster

A High Performance Computing (HPC) based fraud detection system that uses a virtual cluster to process financial transactions in parallel using machine learning.

Features

• HPC cluster with Headnode and Compute Nodes
• PXE Boot provisioning
• LDAP-based centralized authentication
• NFS shared storage
• SLURM job scheduling
• XGBoost fraud detection model
• PostgreSQL database
• Grafana monitoring dashboard

Tech Stack

• Ubuntu Server 24.04
• VMware Workstation
• SLURM
• LDAP
• NFS
• Python
• XGBoost
• PostgreSQL
• Grafana

Architecture

Transactions
      │
      ▼
 Headnode (SLURM)
      │
      ▼
 Compute Nodes
      │
      ▼
 ML Fraud Detection
      │
      ▼
 PostgreSQL
      │
      ▼
 Grafana Dashboard

Current Status

• ✅ Virtual HPC Cluster Setup
• ✅ PXE Boot Configuration
• ✅ LDAP Server Setup
• ✅ NFS & SLURM Configuration
• ✅ Fraud Detection Pipeline
• ✅ Dashboard & Monitoring

---

Author: Arunima Mukhopadhyay
Course: PG-DAC HPCSA, C-DAC Pune