What you can do here

AI Academy is a local AI and DevSecOps learning project. It combines an Ollama model server, a FastAPI API, a document retrieval store, and command-line workflows. This site explains the checked-in implementation; it does not certify a running installation.

Choose your next step

Your situation	Start here
Installing for the first time	Requirements
The backend is already running	Your first query
Using the installed /opt/devsecops-ai/scripts tools	ask.sh, devai.sh, and plan_exec.sh
Building an application or extending the backend	Build and develop
Learning by doing	Hands-on labs
Something is broken	Troubleshoot a problem

Capability status

Capability	Status in this checkout
Local inference through Ollama	Implemented; requires downloaded models
Routing between the two configured models	Implemented; availability can override selection
Retrieval from ingested documents	Implemented using embedded ChromaDB
Conversation history	Implemented with isolation and retention limitations
Prompt/playbook catalogs and planner handoffs	Implemented workflow assets; human review remains necessary
Static documentation on GitLab Pages	Implemented build pipeline
QLoRA model training	Experimental; quality improvement is not established
Docker and automated service deployment	Configuration exists; known build/startup issues remain
Kubernetes, SSO, multi-tenancy, production hardening	Not established as supported deployments

Local does not mean automatically private

Inference through your local Ollama server stays on that path. Model downloads require network access, and hosted assistants introduce a different data boundary. Logs, backups, and shared conversation storage also need protection. Use non-sensitive sample data until you have reviewed the security model.

Next: Check your environment.