Skip to main content
🎤 Luca Berton is speaking at Red Hat Summit & KubeCon EU 2026!Learn more →
intermediate

Observability for AI Workloads

Follow a request through a Kubernetes AI service and use traces, metrics, and logs to diagnose model and retrieval problems.

6 lessons21mUpdated November 2026

What you'll learn

  • Measure time to first token, throughput, and queue pressure
  • Collect OpenTelemetry signals and correlate traces with logs
  • Choose indicators and route actionable alerts

Who this is for

  • • DevOps engineers deepening their expertise
  • • SREs building production-grade systems
  • • Backend developers expanding into infrastructure

Prerequisites

  • • A Mac that can run a local Kubernetes cluster, and familiarity with kubectl.

Course content

2 modules, 6 lessons, 21m of video

Module 1

3 lessons, 11 min
1

4 min

Time to first token, tokens per second, queue depth, errors, accelerator memory and cost per request.

2
The collection stack

4 min

Prometheus scrapes the model servers; the OpenTelemetry Collector carries metrics, traces and logs.

3
Indicator governance

4 min

A catalog of indicators, and a check that fails a change which breaks it.

Module 2

3 lessons, 10 min
4
Alert routing

5 min

Alertmanager routes per team and severity, inhibition for symptoms, silences for planned work.

5
Logs correlated by request id

4 min

Loki, the collector reading pod logs, and one request id that joins them.

6
Distributed tracing

3 min

Spans, the traceparent header, and Tempo: find the slow step in one look.

Get started

Start Free Lesson

First lesson is always free — no signup required

This course includes

6 hands-on lessons
Intermediate level
21m total
Final assessment

Want all courses?

See Pro plan →

Your instructor

Luca Berton

Founder and CEO, Open Empower. About the author

More in AI Agents

All courses in this track