AI Campus • Practical resources
Prompt Engineering & System Architecture
Turn a task into clear instructions, examples and testable output requirements.
Before you start
- A concrete task and ten sample inputs; no paid model account is needed to write and review prompts.
The reading and planning exercises need no paid course or API. Optional inference, training, IDE features and hosted services may require hardware, accounts or paid access. Use the local or manual alternative before spending money.
Start with the official learning material
Free-to-read references checked 3 November 2026. Reading documentation does not activate a provider or create a subscription.
Practice with a concrete exercise
1. Define success first
Read the overview. Choose a task such as extracting event title/date from a notice. Write the required fields, an unknown-value convention and three examples of acceptable output before writing the prompt.
2. Create two prompt versions
Write a baseline instruction and a revision that separates task, input boundaries, output schema and examples. Ask for a concise result or explanation, not hidden reasoning. Include a notice that contains an instruction to ignore the task.
3. Review on fixed cases
Manually simulate or use an already available local model on ten cases, keeping inputs identical between prompt versions. Validate JSON and record missing fields and invented dates. No model access? Peer-review the prompts and label the result a design review, not a model evaluation.
Check what you learned
- Missing information stays unknown rather than being invented.
- Outputs are checked against the declared schema.
- Any claimed improvement includes actual baseline/candidate evidence.
Keep your artifact, observed checks and remaining limitations in a short README. Mark any exercise you only designed rather than executed.