Uploaded August 2026 | Updated September 2026, 2 days ago
In this solo AMA livestream, Viktor tackles a wide range of questions spanning infrastructure provisioning, AI tooling, and career advice. He shares his strong preference for Crossplane over Terraform, Ansible, and CloudFormation for managing multi-cluster Kubernetes environments, explaining the unique value of Crossplane compositions for building custom APIs. He also weighs in on Cluster API, immutable infrastructure with Talos, managed Kubernetes services like EKS, AKS, and GKE, and why he avoids monorepos — especially in an AI-driven development workflow where smaller, focused codebases help agents perform better.
A significant portion of the session focuses on AI agents and how they're reshaping the way engineers work. Viktor discusses integrating AI into CI/CD pipelines — arguing that AI should fix failing tests rather than just explain them — and shares his personal setup using a swarm of autonomous coding agents coordinated through his open-source project, Agent Deck. He also touches on running local LLM inference in air-gapped environments, the evolving role of internal developer portals like Backstage in an agent-first world, and why he believes technical management skills are the most valuable thing engineers can develop as AI takes over more execution work.
▬▬▬▬▬▬ ⏱ Timecodes ⏱ ▬▬▬▬▬▬
00:00 Intro (skip to first question)
06:29 Multicluster provisioning: Crossplane vs Terraform vs CloudFormation
09:16 Do you use OpenAPI spec or AI spec frameworks?
10:24 What do you think about VKS?
11:22 Cluster API vs Crossplane for cluster provisioning
13:59 Immutable infrastructure: good idea or not?
16:07 Advice for starting DevOps career in your 40s
18:51 How to stay valuable as DevOps engineer in AI-driven world
21:19 When to use object storage vs NFS for Kubernetes apps
22:30 How to differentiate yourself when PMs also use AI
24:20 Home lab vs managed Kubernetes for learning
26:41 LLM architecture for air-gapped environments
28:48 Running VLMs on Kubernetes with cheap GPUs
30:44 Integrating AI inference with IDP like Backstage
33:19 How to implement AI in CI/CD pipelines
36:46 Is Kubernetes a natural fit for AI workloads
38:40 DeepSeek Harness vs OpenHands for software engineering
41:31 AI for analyzing failed Jenkins pipelines and pod logs
46:37 Mono repo vs poly repo for modern development
48:35 Alternatives to Anthropic SDK for building agents
51:08 Is AI replacing DevOps roles
53:05 AWS certifications: worth it or not
57:37 AKS and Bicep: is it a good option
01:02:59 How to make agents remember context across sessions
In this solo AMA livestream, Viktor tackles a wide range of questions spanning infrastructure provisioning, AI tooling, and career advice. He shares his strong preference for Crossplane over Terraform, Ansible, and CloudFormation for managing multi-cluster Kubernetes environments, explaining the unique value of Crossplane compositions for building custom APIs. He also weighs in on Cluster API, immutable infrastructure with Talos, managed Kubernetes services like EKS, AKS, and GKE, and why he avoids monorepos — especially in an AI-driven development workflow where smaller, focused codebases help agents perform better.
A significant portion of the session focuses on AI agents and how they're reshaping the way engineers work. Viktor discusses integrating AI into CI/CD pipelines — arguing that AI should fix failing tests rather than just explain them — and shares his personal setup using a swarm of autonomous coding agents coordinated through his open-source project, Agent Deck. He also touches on running local LLM inference in air-gapped environments, the evolving role of internal developer portals like Backstage in an agent-first world, and why he believes technical management skills are the most valuable thing engineers can develop as AI takes over more execution work.
▬▬▬▬▬▬ ⏱ Timecodes ⏱ ▬▬▬▬▬▬
00:00 Intro (skip to first question)
06:29 Multicluster provisioning: Crossplane vs Terraform vs CloudFormation
09:16 Do you use OpenAPI spec or AI spec frameworks?
10:24 What do you think about VKS?
11:22 Cluster API vs Crossplane for cluster provisioning
13:59 Immutable infrastructure: good idea or not?
16:07 Advice for starting DevOps career in your 40s
18:51 How to stay valuable as DevOps engineer in AI-driven world
21:19 When to use object storage vs NFS for Kubernetes apps
22:30 How to differentiate yourself when PMs also use AI
24:20 Home lab vs managed Kubernetes for learning
26:41 LLM architecture for air-gapped environments
28:48 Running VLMs on Kubernetes with cheap GPUs
30:44 Integrating AI inference with IDP like Backstage
33:19 How to implement AI in CI/CD pipelines
36:46 Is Kubernetes a natural fit for AI workloads
38:40 DeepSeek Harness vs OpenHands for software engineering
41:31 AI for analyzing failed Jenkins pipelines and pod logs
46:37 Mono repo vs poly repo for modern development
48:35 Alternatives to Anthropic SDK for building agents
51:08 Is AI replacing DevOps roles
53:05 AWS certifications: worth it or not
57:37 AKS and Bicep: is it a good option
01:02:59 How to make agents remember context across sessions
