Keep some work that records an experiment, not only the pieces that prove competence.

A portfolio pruned to its successes becomes a record of what you already knew how to do. The failed attempts hold the actual information: where your edge sat that year, what problem you were circling, how far off the first guess landed. Competence is easy to demonstrate and dull to look back on.

Keep one folder that is not for showing anyone. Put the experiments in it, including the ones that went nowhere, and date them so the movement is visible later.