Learn by doing

Start with a working lab, then practice one capability at a time. You do not need to complete an entire DevSecOps curriculum before asking your first local AI question.

Choose your path

Path	Sequence	Completion evidence
New to the platform	Requirements → install → verify → first query	A nonempty API answer and recorded model name
Local AI and retrieval	API lab → routing lab → RAG lab	Explain selection and show retrieved source content
AI-assisted engineering	Framework → planner handoff → assistant workflow	A reviewed plan with checkable acceptance criteria
Build and develop applications	API client → backend extension → prompt evaluation → review	A working client and a tested change in a development copy
Platform operations	Services → logs → backup/restore → updates	A recorded recovery rehearsal in an isolated environment

Start with installation or go directly to the lab catalog.

Prerequisites and scope

Each lab states its prerequisites and expected result. Timing depends on model downloads and inference speed. The site does not provide automated progress tracking or certification.

Progress from foundations to engineering

The wider Academy curriculum remains relevant. Use this progression to decide what to learn next; it does not require every topic before using the local AI lab.

Stage	Skills to develop	A concrete learning artifact
Systems foundation	Windows versus WSL terminals, files, permissions, processes, ports, and logs	Explain where your backend runs and where it writes data
Git and GitLab	Commits, branches, diffs, merge requests, and pipeline results	A small reviewed documentation change
Local AI	Model serving, context length, routing, retrieval, and evidence	The three core labs
Application development	HTTP clients, JSON contracts, validation, error handling, and tests	The client and endpoint exercises in Build and develop
Infrastructure and automation	Read Ansible, Terraform, containers, and CI configuration	A reviewed sample configuration with its assumptions and validation plan
DevSecOps	Secret handling, least privilege, threat review, and release controls	A scoped security assessment tied to actual code and configuration
AI-assisted engineering	Prompt design, handoffs, code review, and evaluation	A concrete plan, implementation diff, and verification record

Apply AI to DevSecOps learning

Choose one small problem and supply the relevant sample code. Ask the assistant to explain it, propose a change, and identify how to verify the result. Useful projects include:

• Ansible: explain a local sample playbook, then propose an idempotent improvement. Distinguish generating YAML from actually configuring a machine.
• Terraform: review a sample resource’s inputs, outputs, and state implications. Validate configuration before considering any infrastructure action.
• GitLab CI: draft a job for a deterministic check and inspect its rules and failure behavior. A YAML example is not proof that the job ran.
• Knowledge assistant: ingest your own non-sensitive examples and return supporting sources with each answer.
• Review helper: build on the API client to submit a small diff and present findings for human review.

These are guided extension projects, not preinstalled capabilities or certified deployment recipes. The core API/routing/RAG labs have dedicated steps; broader infrastructure and cloud projects require their own environment and task-specific validation.

Keep a learning record

For each exercise, keep the goal, source revision, sample input, observed output, checks performed, and what you would improve. Move on when you can explain the outcome and reproduce it, rather than after a prescribed number of hours.

Experimental learning

Model training documents the existing distillation experiment, what its artifacts establish, and what remains to evaluate.

Next: Choose a lab.