17 free courses — no signup wall
Architect-led enterprise cloud, security & AI
320+ downloadable toolkits — instant delivery
Skip to content
Platform engineer

AI Platform Engineering

The runtime every model lands on: GPUs, serving, autoscaling, and the bill at the end of it.

Who this is for

Platform engineers who own the paved road other teams deploy models onto.

What you should be able to do

Serve a model with autoscaling, observability, and a cost ceiling that holds.

Career ladder

The titles this track maps onto. Levels differ between companies — the useful part is the direction, and what each step adds to the one before it.

  1. Platform Engineer
  2. AI Platform Engineer
  3. Senior Platform Engineer
  4. Staff Platform Engineer
  5. Principal Platform Architect

Tech stack

What the work is actually done with. Grouped by the job each tool does, so the list reads as a system rather than a pile of names.

Compute
KubernetesGPU node poolsKarpenterRay
Serving
vLLMRay ServeKServeBedrockVertex AI
Observability
OpenTelemetryPrometheusGrafana
Cost
FinOpsRight-sizingSpot / preemptibleToken budgets

The delivery flow

The order the work actually happens in. Each step is where a decision gets made and written down, not a chapter heading.

  1. Capacity plan
  2. Cluster + GPU pool
  3. Model serving
  4. Autoscaling policy
  5. Telemetry
  6. Cost guardrails
  7. Multi-tenancy
  8. Paved-road docs

Reference repository structure

A starting layout for this track. The directories are the ones that get added late and hurt — decisions, evals, policy, lineage — promoted to the top level where they are visible.

ai-platform/
├── clusters/              # one directory per environment
│   ├── gpu-pools/
│   └── autoscaling/
├── serving/
│   ├── vllm/
│   └── kserve/
├── observability/
│   ├── otel-collector/
│   ├── dashboards/
│   └── alerts/            # SLOs, not just thresholds
├── cost/
│   ├── budgets/
│   └── rightsizing/
├── tenancy/               # quotas and isolation per team
└── paved-road/            # the docs that make self-service work

Reference implementations

Citadel’s open-source repositories for this track — Terraform modules, MCP servers, and reference architectures you can read, fork, and run. Apache/MIT licensed; check each repository for its terms.

Primary sources

The standards, framework documents, and vendor references this track is built against. Go here when you need the authoritative wording rather than a summary — in a security review or an audit, the source is what counts.

Other tracks

Talk through this with an architect