2 min read
Training
Free observability training with lessons, reproducible labs and action plans, organized by theme: artificial intelligence, IT and infrastructure, industry, governance, open source stacks.
Every MTTL course follows the same thread. You start by framing the problem, often from an incident or a question asked by leadership. You then move to practice, pick what is worth measuring and finish with an action plan for running it at scale. A lesson reads in one sitting.
Some courses include labs: small environments you start on your own machine, reproducing a real instrumented stack. Each one starts with docker compose up. Others, such as “Understanding HPC” or the CIO path, need no code at all.
Your progress
Section titled “Your progress”Your progress is stored in this browser and nowhere else. Every multi-lesson course ends with a quiz; programmes still in preparation do not have one yet.
- Understanding generative AI observability
- LLM observability: the labs
- GenAI observability method
- Understanding high performance computing
- HPC AI: observability and security reference architecture
- Governing observability: the CIO path
- VictoriaMetrics as an LLM observability backend
A lesson counts as read once you have scrolled through 80% of it, kept it open for more than 30 seconds, or ticked “Mark as read”. This information stays in your browser.
Artificial intelligence
Section titled “Artificial intelligence”Observing LLMs, agents and RAG systems in production: answer quality, drift, cost and regulatory evidence. The courses follow a progression: understand (the guide), practise (the labs), master (the method), then go beyond the LLM.
- PractitionerUnderstanding generative AI observabilityThe reference guide, to read first: LLMs, RAG, agents and MCP; five signals, four system types, six maturity levels, seven steps, evaluators, tools, operations.
- PractitionerLLM observability: the labsHands-on, after the guide: 5 modules and 4 labs on a sovereign stack for LLMs and RAG, GenAI conventions, drift, continuous evaluation, cost and SLOs.
- ExpertGenAI observability methodTo master the subject, once the guide and the labs are done. LLM, agent, RAG and MCP: silent failure, quality SLOs with the Wilson bound, quantified impact with its calculator, regulatory evidence, RACI.
- In preparationBeyond the LLM: GPUs and model qualityGoing further, assuming the LLM course: GPU infrastructure, classic ML, statistical drift detection, retraining.
IT and infrastructure
Section titled “IT and infrastructure”Understanding, observing and securing compute infrastructure: the HPC clusters and GPUs AI is trained on.
- BeginnerUnderstanding high performance computing3 lessons, no code. What an HPC cluster is, how a job runs on it and why large-scale AI depends on it.
- PractitionerHPC AI: observability and security reference architectureThe five layers of a training cluster, anatomy of a step, the observability stack, eBPF and six attack vectors.
- In preparationHPC observability foundationsRedfish exporter, DCGM and Slurm join, MIG contention, a director-readable dashboard.
Industry
Section titled “Industry”Observing industrial IT: controllers, field networks, fleets of connected devices and resource-constrained sites.
No course has been published on this theme yet. The “Edge and IoT” chapter is planned: see the roadmap.
Governance and steering
Section titled “Governance and steering”For decision makers: funding, governing, complying and steering observability.
Open source stacks and tools
Section titled “Open source stacks and tools”Getting hands-on with one component of the observability stack, from installation to operations.
Where to start?
Section titled “Where to start?”For generative AI, the content forms a six-stage progression. You discover with AI in cross-section, understand with the guide, practise with the LLM labs. You then go to production with the playbooks, then VictoriaMetrics for LLMs. You master the subject with the GenAI method, before going beyond with GPUs and model quality. Depending on your role, you join at a different point:
| You are | Start with | Then |
|---|---|---|
| Platform engineer or SRE new to AI | AI in cross-section, then Understanding generative AI observability | The LLM labs, then the deployment playbooks and VictoriaMetrics for LLMs |
| ML engineer or MLOps | Understanding generative AI observability, then the LLM labs | GenAI method, then Beyond the LLM |
| Architect | AI in cross-section, then Understanding generative AI observability | Deployment playbooks, VictoriaMetrics for LLMs, GenAI method; HPC AI if you run GPUs |
| HPC administrator | HPC AI: reference architecture | HPC observability |
| CIO, leadership, finance | CIO path | How much does an unmonitored AI cost?, GenAI method parts V to VII |
| Team manager | CIO path, lessons 3 and 5 | Too many alerts: is on-call sustainable? |
If you are unsure, I recommend starting with the shortest course in your theme: you will quickly see whether the level suits you.
Revised on 4 October 2026: “Where to start” table aligned with the progression of the generative AI content (guide before the method, labs, production, beyond the LLM).