Enabling Multi-Cloud Architecture for a Zurich Fintech, Combining AWS and GCP

Client Profile

A Zurich-based fintech company providing payment processing and fraud detection services to European banks and online retailers. The team of 120 engineers and analysts serves clients across Switzerland, Germany, and Austria.

Industry Financial Technology
Location Zurich, Switzerland
Company Size ~120 employees
Duration 6 months

Technologies Used

AWS Google Cloud Terraform Kubernetes

Business Challenge

The client ran their entire stack on AWS but needed Google Cloud’s Vertex AI platform for a new real-time fraud detection model. Their compliance team also required multi-cloud redundancy to satisfy banking partner requirements around business continuity. The challenge was integrating two cloud providers without introducing operational complexity or latency in the payment processing pipeline.

Solution

We designed a multi-cloud architecture with AWS handling core transaction processing and data storage, and GCP running the ML inference pipeline. Terraform managed infrastructure across both providers from a single codebase. Kubernetes clusters on both EKS and GKE provided deployment consistency. We established secure, low-latency cross-cloud networking via dedicated interconnects and implemented unified monitoring through Grafana Cloud.

Outcome

The fraud detection model deployed on GCP reduced false positive rates by 35% compared to the previous rules-based system. Cross-cloud latency remained below 15ms for inference calls. The multi-cloud architecture satisfied banking partner continuity requirements, and the unified Terraform codebase allowed the ops team to manage both clouds without doubling their workload.

Process

1

Requirements and Compliance Review

Mapped banking partner requirements for business continuity, data residency, and provider redundancy. Defined the technical boundaries between AWS and GCP workloads.

2

Network Architecture

Established secure cross-cloud connectivity with dedicated interconnects, ensuring sub-20ms latency between AWS and GCP for real-time inference calls.

3

Unified Infrastructure as Code

Built a single Terraform codebase managing resources across both AWS and GCP, enabling the ops team to provision and update infrastructure consistently.

4

ML Pipeline Deployment

Deployed the fraud detection model on GKE with auto-scaling inference endpoints, integrated with the AWS-hosted transaction pipeline via secure API gateway.

5

Monitoring and Observability

Configured Grafana Cloud for unified dashboards across both providers, with PagerDuty alerting on cross-cloud latency thresholds and service health.

6

Failover Testing

Conducted controlled failover exercises to validate business continuity across providers, documented procedures for the client's compliance reporting.

Conclusion

The multi-cloud architecture delivered the best of both providers — AWS for reliable transaction processing and GCP for advanced ML capabilities — while meeting the strict compliance and continuity requirements of the financial services industry.

Ready to Transform Your Infrastructure?

Book a free consultation with our team to discuss your DevOps and cloud engineering needs.