OSPI Certified: Ollama Administrator

Exam objectives

1. 1. Linux host readiness for local LLM workloads
   • Inspect CPU, memory, disk space, and GPU availability using standard Linux tooling
   • Relate host capacity to model storage and inference requirements
   • Record resource findings clearly when asked to document the environment
2. 2. Ollama installation, service management, and API basics
   • Install Ollama on Linux using supported official methods
   • Manage the Ollama daemon with systemd (systemctl)
   • Configure service environment variables on the Ollama unit
   • Work with the Ollama HTTP API: understand how it is exposed and verify it responds
3. 3. Model selection and deployment
   • Compare model sizes and choose an appropriate tag for available disk and RAM
   • Pull, run, and remove models with the Ollama CLI
   • Send prompts and confirm the model produces expected responses (CLI or API)
4. 4. Interactive CLI and session management
   • Work in an interactive Ollama session and use built-in session commands
   • Adjust runtime options (such as parameters or system context) and inspect current settings
   • Save and resume conversational sessions when supported
   • Inventory installed models from the command line
5. 5. Model customization and advanced capabilities
   • Read and author a Modelfile (base model, parameters, system message, templates)
   • Build a custom model tag from a Modelfile and validate it is available locally
   • Understand which model families support extended capabilities documented by Ollama (e.g. reasoning/thinking modes)
   • Run capability-enabled models and interpret their output format

   Study: Create a customized model, Thinking, and related guides on docs.ollama.com

6. 6. Operations, logs, and troubleshooting
   • Find and read Ollama service logs using journalctl
   • Diagnose typical operational failures: service not running, failed model pull, unreachable API
   • Apply a logical remediation path and verify the service and API are healthy afterward