Stingray is a crypto-intelligence platform providing AI-powered analytics, real-time alerts, and social publishing tools. It was running on GCP, but Vertex AI's model restrictions meant the team couldn't access the model infrastructure they needed. We migrated the full 8-microservice stack from GCP to AWS, re-platforming onto ECS Fargate with Terraform-managed infrastructure, and handed over a production-ready multi-account environment.
Stingray's GCP stack was functioning, but model access was the blocker. Vertex AI's model restrictions limited access to the broader model infrastructure the team needed. AWS Bedrock provided the model access they needed.
The challenge was migrating a live platform, 8 microservices in production, without disrupting active development, and building a new AWS environment that would be production-grade from day one.
Multi-account AWS Organization
We built the AWS environment as a proper multi-account organization from the start — separate accounts for production, staging, dev, security, and tooling. A read-only assessment of Stingray's GCP infrastructure and codebase came first, producing a detailed migration plan before a line of Terraform was written.
ECS Fargate and CI/CD
All 8 microservices were containerized, pushed to ECR, and deployed as ECS Fargate tasks behind an Application Load Balancer. A three-tier GitHub Actions pipeline handles deployments - dev on demand, staging on merge, production as a manual promotion - authenticating via OIDC with no long-lived credentials in the repository.
Security and access
Centralised access management provides a single source of truth across the entire organization. A security layer covers continuous compliance monitoring and vulnerability detection. All infrastructure is managed exclusively via Terraform.
Data migration and cutover
All database tables were migrated from GCP Cloud SQL to RDS PostgreSQL. Embeddings were regenerated using the Titan model rather than migrated directly. The final production cutover ran in a 1-hour maintenance window: database dump from GCP, restore to AWS, DNS records switched.
"Apex Lab helped us migrate a complex AI product from GCP to AWS without disrupting product development. They gave us a solid Terraform-managed AWS foundation and handled the cutover cleanly. The result was a scalable infrastructure setup with access to the model capabilities we needed."
Stingray's full 8-microservice platform is migrated from GCP to a production-ready AWS environment. Every resource is managed via Terraform. Development continued on GCP throughout the build phase with no disruption, and the production cutover completed within the approved 1-hour window. Stingray now has broader access to model infrastructure, including Claude through AWS Bedrock, on an infrastructure foundation built for scale.
Tell us more about yourself and what you're got in mind.
Prefer email? hello@apexlab.io
Or even a quick call?
We don’t care about fancy CVs or long interview processes. Provide some basic information, let’s have a quick chat, and then show us what you love doing the most: write some code and create a product.
