Kubernetes
Named by 26 kept items in this workspace.
14 items name it under frameworks and tools and 12 under classification.technologies.
The items that name it 26
Container orchestration tool; Bernhardsson says it and Docker were built for steady backend workloads, not GPU-heavy gen AI workloads
Prior-generation container orchestration tool; Erik says it and Docker were built for steady backend workloads, not expensive GPUs and large models
Repository (31k files) used as the speed benchmark target
Open source system, developed by Google, for deploying and managing cloud applications at scale; uses control loops to keep applications running, recover from failures, and scale when needed; kickstarted the "cloud native" era of computing.
Orchestrator tested for 5 days per provider, including pod churn, sandbox pod batches and PVC lifecycle.
Autoscaling to scale out the gateway
Orchestrates per-agent pods for Cumora Cloud agents, controlled via kubectl from the server.
Target platform managed by AliangBoard, covering multi-cluster management, exec, port-forward and debug-container injection
Container orchestration for Hugging Face's production environment, which the agent penetrated and where it deployed a self-respawning pod fleet across 11 nodes.
Orchestration layer the service is deployed on, with autoscaling driven by queue depth rather than CPU
- Training State of the Art Vulnerability Discovery Agents through Reinforcement Learning | depthfirst
Used to spin up and run thousands of sandbox environments for RL training, where the agent audits a repository using only a shell.
Infrastructure where AI agents are deployed as workloads
Platform for deploying the OpenShell gateway; the CNI must enforce NetworkPolicy
Cluster platform that AX runs on top of and feels similar to; ax follows the active Kubernetes context to resolve and tunnel to that cluster's control plane.
In every workspace you can read 115
Security
AI
All technologies in this workspace.