What criteria should you use to choose a project? Return JSON {"criteria": [list of 2+ criteria], "why": "short"}.
A portfolio project is needed to prove a skill, not to impress with its scale. Best choice: solves a real, understandable problem (even a small one); FINISHED and works from beginning to end - the modest completed is stronger than the unfinished grandiose; shows the required skill clearly (if you are targeting AI engineers, show RAG, agent, integration); can be explained and demonstrated in a couple of minutes. One neat completed case is more convincing than five abandoned ones. Rule: take a narrow, realistic task and bring it to a working conclusion; “ready and working” is more valuable than “ambitious but not finished.”