Handling canary deployments with GitHub Actions and ArgoCD

In this session, I will walk you through how we at Liftoff Mobile successfully designed and implemented a self-service production deployment pipeline that enables engineering teams to seamlessly deploy services across multiple Kubernetes clusters—without causing service disruption, even while handling a sustained 300,000 requests per second (per cluster).

I will break down the key aspects of our Kubernetes architecture, the challenges we faced, and the lessons we learned while building a deployment process that ensures high availability, reliability, and developer autonomy. This session will provide an in-depth look into the tooling and automation that make this possible, including:
- JIRA for issue tracking and release coordination
- GitHub Actions for CI/CD automation
- Argo Rollouts for progressive delivery and canary deployments
- ArgoCD for declarative GitOps-based cluster management
- Kubernetes (EKS) as our cloud-native platform

Beyond the technical implementation, I will also touch on the business and operational drivers behind this initiative—why we needed to overhaul our deployment process, what challenges we faced before this transition, and how we navigated the change to empower engineering teams with a fully automated and self-service approach to production deployments.

This session is for engineering leaders, DevOps teams, and platform engineers looking to scale Kubernetes deployments efficiently while maintaining reliability, developer autonomy, and operational excellence in a high-traffic production environment.