AI Infrastructure Capstones
Every AI Infrastructure Capstones lesson in the Linux for AI Engineering course, in the order you study them. Each page opens about the first half of its lesson to read, some the whole of it; the full lessons are in the plan.
Capstone A: a private AI workstation, from bare metal to a sandboxed agent
Capstone: build a Linux AI workstation in layers, from driver to services, and survive the kernel update that breaks the NVIDIA module.
Capstone B: from dataset to fine-tune to a monitored production endpoint
Capstone: a pipeline from data to fine-tune to production endpoint, and why a model great in testing behaves strangely when served.
Capstone C: a GPU Kubernetes inference platform built as code, with autoscaling
Capstone: a GPU inference platform on Kubernetes with a gateway, metrics and autoscaling, and why scaling on CPU fails for LLMs.
Capstone D: an HPC training cluster with InfiniBand, shared storage and Slurm
Capstone: build an HPC GPU cluster with Slurm, InfiniBand and shared storage, and catch NCCL quietly falling back to TCP.
Capstone E: AI infrastructure system-design interviews
AI infrastructure interview questions and drills: the shape of a good design answer, numbers to know, and sizing a serving platform.
