AI Infrastructure

AI/ML Engineering Track | Phase 6

Best for: learners moving from model/application work into inference, cost, and infrastructure decisions.

This phase is intentionally split between:

• learner-scale local infrastructure
• production-oriented AI infra concepts

That means not every module should be read in pure numeric order if your immediate goal is local-first work.

Modules

#	Module
1.1	Cloud AI Services
1.2	AIOps
1.3	High-Performance LLM Inference: vLLM and sglang
1.4	Local Inference Stack for Learners
1.5	Home AI Operations and Cost Model
1.6	GPU Memory Hierarchy and Bandwidth Math for LLM Inference
1.7	Production-Tier LLM Inference Engines: Decision Framework
1.8	Benchmarking LLM Inference: TTFT, TPOT, and Workload-Aware Load Shaping

Suggested Paths

Local-First Route

• Local Inference Stack for Learners
• Home AI Operations and Cost Model
• then branch into Single-GPU Local Fine-Tuning or Home-Scale RAG Systems

Production-Oriented Route

• Cloud AI Services
• High-Performance LLM Inference: vLLM and sglang
• then continue into Platform Engineering: Data & AI or On-Premises AI/ML Infrastructure

Key Distinction

This section is not only about datacenter-scale AI. It also teaches when a learner should stay simple, local, and private instead of prematurely copying large-scale serving architecture.