Modernizing Kubernetes delivery
Dec 12, 2025
In this interview, Amine Hilaly, Software Development Engineer at AWS, discusses:
GitOps-driven infrastructure management: How organizations use Argo CD combined with AWS Controllers for Kubernetes (ACK) to deploy entire infrastructures, including EKS clusters themselves, through centralized management clusters
Scaling beyond kubectl for multi-cluster operations: Strategies for maintaining operational consistency across dozens of clusters by embracing standardization and automated tooling rather than direct cluster management
Transcription
Bart: So, who are you, what's your role, and where do you work?
Amine: Hi, my name is Amine. I'm a software engineer at AWS, working specifically in the EKS team within the Kubernetes ecosystem. I'm based in Seattle.
Bart: Fantastic. What are three emerging Kubernetes tools that you're keeping an eye on?
Amine: I would say Karpenter, Bootc, and Kro.
Bart: Platform teams spend significant time building and maintaining deployment pipelines. How should organizations think about modernizing their Kubernetes application delivery?
Amine: One of the ways to do that is to embrace new tools and emergent tools in the Kubernetes ecosystem. We always try to innovate and invent new tools and bring new ways of solving these problems. Modernizing your pipelines means embracing new tools, designs, and architectures behind the scenes.
Bart: Tools like Argo CD, ACK, and KRO are gaining significant traction in the Kubernetes ecosystem. How do these tools fit into AWS's vision for container platforms?
Amine: I think those are the most used tools in EKS. A lot of customers use them. Many embrace the GitOps methodology with Argo CD. They love to deploy everything to Kubernetes instead of having to split their focus or use multiple tools to solve the same problem. For example, we have many customers using Argo CD in combination with ACK to deploy their full infrastructure behind the scenes. In fact, one of the biggest use cases is deploying EKS clusters using ACK and GitOps. We have a management cluster, a centralized cluster that provisions other clusters for internal platform teams.
Bart: As organizations grow from a few clusters to dozens across multiple regions, what strategies help them maintain operational consistency without building complex custom solutions?
Amine: I think standardization is definitely one of the ways to do that. At some scale, starting with 10 clusters in production, what you need is a process and mechanism that helps you manage multiple clusters and allows your team to scale beyond managing just one cluster. One way of dealing with that is to stop using kubectl. Instead, embrace tools like Argo CD, ACK, or Kro that can manage access to other clusters for you, rather than managing them directly.
Bart: What's your biggest learning from this KubeCon?
Amine: I learned about a new concept called gang scheduling. There's a concept in Kubernetes about scheduling multiple pods simultaneously. I've been talking to many folks at KubeCon about gang scheduling, exploring its applications and how we're trying to solve the problem differently.
Bart: How can people get in touch with you?
Amine: I'm available on LinkedIn, Twitter, and Blue Sky. I'm excited to chat.
