Context
Koo’s production platform consisted of dozens of application and ML services running on AWS, serving a large active user base across regions. The program aimed to migrate the full production stack to GCP within a strict four-month timeline.
My Role
Lead DevOps Engineer owning end-to-end migration execution, cloud architecture, infrastructure automation, and production cutover planning.
Challenges
- Aggressive four-month delivery timeline
- Large number of interdependent services and databases
- Active production traffic during migration
- Zero-downtime requirement for user-facing systems
What I Did
- Designed and implemented a production-grade GCP landing zone
- Migrated 40+ application services, 48 ML services, and multiple databases
- Planned phased service cutovers with clear rollback strategies
- Automated infrastructure provisioning using Terraform
- Standardized CI/CD pipelines using Jenkins
- Defined runbooks, cutover checklists, and governance processes
Outcome
- Zero unplanned downtime during production migration
- Migration delivered within the committed four-month timeline
- Improved platform performance and operational efficiency
- Reduced long-term infrastructure and operational costs