AI engineering portfolio: make the work inspectable

What to inspect

Describe the task, your contribution, the system boundary and what the evidence supports. Include a runnable example or artifact, one failure and the check used to detect it. Make production status and unresolved limitations explicit.

Worked example

This library’s invoice example is a teaching fixture with local checks. It can demonstrate validation logic. It cannot establish model accuracy, production scale, client savings or employment experience. A credible case study preserves that distinction.

Copy the full prompt

Adapt the inputs to your task. Remove private data before sending anything to a model. This page copies text locally; it does not run the prompt.

Turn [PROJECT NOTES] into a technical case study. Sections: problem, my verified contribution, architecture, setup, input/output example, failure case, checks run, results, limitations and next steps. Cite actual artifacts. Mark unverified claims for removal or follow-up. Do not invent customers, deployment status, savings or responsibilities.

Review checklist

1. State your own contribution accurately.
2. Link to a runnable example or inspectable artifact.
3. Show one failure and how it is detected.
4. Separate prototype status from production evidence.

Keep the resource

Keep exploring

• LLM evaluation: measure the usable result
• AI evaluation cases: turn a failure into a regression test

Documentation & scope

Anthropic: harnesses and verification ↗

This is an original workflow template with an illustrative example. It does not report completed model tests or measured performance. Provider APIs can change; check the linked documentation for your exact integration.