Hands-on builds for the AI infrastructure substrate, from GPU allocation to the agentic layer.
- AI Native Hub - Entry point: running AI and ML workloads on Kubernetes, indexed by domain
- Dynamic Resource Allocation - Declarative GPU tier fallback with DRA, runnable on Kind
- Gateway API Inference Extension - Model-aware request routing with GAIE and agentgateway, no GPU required
- Model Context Protocol - MCP in practice: connecting agents to systems under governance
This organization hosts the builds behind the writing on how Kubernetes is becoming the substrate the AI stack runs on: accelerator allocation, distributed inference, intelligent routing and the agentic layer forming above them.
Everything here is reproducible. Most labs run on Kind and need no GPU.
The reading is indexed at www.christiandussol.dev.
Maintained by Christian Dussol
All resources in this organization are shared under Creative Commons Attribution-ShareAlike 4.0 International License unless otherwise specified.
