Built from Zero: How a Financial Institution Deployed a Production-Grade Fraud Detection System from Greenfield in Record Time

Client Name : Financial Institution

Industry : Financial Services (AML, Fraud, Sanctions)

Start Date : 02/09/2024

End Date : 08/01/2024

Customer Overview

This Company is an IT solutions provider operating in the US East region. The company specializes in delivering technology-driven services to clients across various industries, including banking.

Business Challenge

The client was tasked with building a brand-new fraud and default detection system for two banking clients, Bank A and Bank B. There was no existing infrastructure in place, and the requirement was to design a solution from scratch.

The system needed to

Accept queries from either bank through a centralized hub.
Route those queries to an orchestration layer for processing.
Query two separate databases—MySQL and SQL Server Express—each containing respective customer data.
Return a decision on whether a user profile was valid or flagged for fraud or default.

This architecture had to be secure, scalable, and capable of handling real-time decision-making across multiple data sources.

Objective

The primary goal of the project was to build a secure, highly available architecture that could support real-time fraud and defaulter detection across banking systems. Client aimed to

Establish a centralized hub and orchestration layer to process queries from multiple banks.

Ensure maximum uptime and fault tolerance using AWS best practices.

Implement a secure environment aligned with AWS Well-Architected Framework pillars, including Security, Reliability, Performance Efficiency, and Operational Excellence.

Solution by Futran Solutions

Futran Solutions delivered a greenfield deployment on AWS, architecting a secure and highly available solution tailored to client’s fraud detection requirements. The design incorporated all pillars of the AWS Well-Architected Framework,

Managed Services – Monitoring & Optimization, Enabled real-time monitoring and alerting using Amazon CloudWatch and AWS Config, ensuring visibility into system health, performance, and compliance.
Managed Services – Patching & Maintenance, Delivered automated patching for Windows Server workloads using AWS Systems Manager Patch Manager, reducing vulnerabilities and ensuring security compliance.
Managed Services – Cost & Compliance Management, Provided ongoing FinOps support via AWS Cost Explorer and Trusted Advisor, helping optimize resource usage and align with financial compliance standards.
Managed Services – Identity Security & Lifecycle Management, Implemented IAM roles, least privilege policies, and VPC segmentation to secure access across orchestration and database layers. Used AWS Config for drift detection and compliance tracking.

Architecture Overview

The solution was deployed in a multi-AZ architecture within a dedicated Amazon VPC, ensuring isolation and security. Key components included:

Public and private subnets for separating application and database layers.
Security groups are tightly configured to allow only necessary traffic between EC2 instances, databases, and external clients.
EC2 instances for the hub and orchestration layers, distributed across Availability Zones for fault tolerance.
Amazon RDS for MySQL and SQL Server Express on EC2 for client-specific data storage.
IAM roles and policies to enforce least privilege access across services.

Security

The solution was designed with security as a foundational principle. IAM roles and policies were implemented to enforce least privilege access across all AWS resources. VPC segmentation and tightly scoped security groups restricted network access to only essential components, reducing the attack surface. All data flows between components were encrypted to ensure confidentiality and integrity. Additionally, AWS Config and Amazon CloudWatch were used to monitor compliance and detect configuration drift, helping maintain a secure and auditable environment.

Reliability

To ensure continuous availability, the architecture leveraged a multi-AZ deployment strategy for both EC2 instances and Amazon RDS. This design allowed the system to remain operational even in the event of an Availability Zone failure. Health checks and automated failover mechanisms were configured to detect and respond to service disruptions quickly. The redundant architecture ensured that critical identity and application services remained accessible at all times.

Performance Efficiency

The system was optimized for performance by selecting appropriate EC2 instance types for the orchestration and hub layers. Database configurations were tuned to support real-time query processing, ensuring low-latency responses for fraud detection workflows. Auto Scaling was implemented to dynamically adjust compute resources based on demand, maintaining consistent performance during peak usage while optimizing resource utilization during off-peak hours.

Operational Excellence

Operational excellence was achieved through centralized logging and monitoring using Amazon CloudWatch, which provided real-time visibility into system health and performance. AWS Config was used to ensure configuration compliance and to track changes across the environment. The architecture was modular, separating the hub, orchestration, and database layers, which simplified maintenance and allowed for easier updates and scaling as business needs evolved.

Key Workloads

The solution was built on a Windows-based stack, leveraging familiar enterprise technologies to meet the needs of the Company’s banking clients. The core workloads included:

Windows Server running on Amazon EC2

IIS (Internet Information Services) for hosting the .NET-based orchestration and hub applications

.NET Framework for application logic and orchestration

SQL Server Express Edition was hosted on EC2 for Bank B’s data

Amazon RDS for MySQL was used for Bank A’s data

Results Achieved

The deployment led to significant improvements in system performance and operational efficiency for the client and its banking clients:

Uptime: Achieved 99.99% availability, with multi-AZ failover reducing downtime risk to <5 minutes per year.

Performance: Reduced query latency by 45%, from ~220ms to ~120ms, enabling near real-time fraud checks.

Operational Impact: Fraud/defaulter detection turnaround time improved by 60%, cutting manual verification effort.

Client Satisfaction: Post-deployment survey from Bank A and Bank B indicated >90% satisfaction with fraud detection accuracy and system responsiveness.

Lessons Learned / Best Practices

Design for High Availability: Multi-AZ deployments and fault-tolerant architecture are essential for mission-critical workloads.
Secure by Default: Implementing IAM roles, security groups, and VPC isolation ensures data protection and compliance.
Modular Architecture: Separating hub, orchestration, and database layers improves scalability and maintainability.
Real-Time Processing: Optimizing EC2 and database configurations enables fast and reliable decision-making.
Monitoring is Key: Using CloudWatch and AWS Config helps maintain operational excellence and proactive issue resolution.

Take Your Business to New Heights

Elevate your business with Futran Solutions' digital transformation. Contact us today to learn more.