Organize by skill
Not by week number only.
article / GVI
Group artifacts by the skill they prove, add context, and revisit the library before applications rather than hunting through chat history.
Not by week number only.
Question and audience on a cover note.
The library is for future you, under time pressure.
00-brief, 01-research, 02-analysis, 03-consumer, 04-recommendation, 05-report-review, 99-certificate. Dated filenames. A README with the one-sentence project.
Artifact, skill, date, whether it is shareable. This index is what you open before an interview. The files can stay messy if the index is not.
Default to private. Share when the data is public and teammates agree. Do not upload other people’s comments with names if they did not expect a public page.
If everything is slides, you never analyzed. If everything is tables, you never recommended. The next GVI module or track should attack the empty folder.
Keep both. Trust the library more. The certificate-versus-portfolio insight is the public version of this argument.
A table: filename, module, skill, date, shareable yes/no, one-line question. Keep it at the top of the library. When an application asks for an example of analysis, you should not have to remember a week number. Future-you is tired. Be kind to them in the present with boring structure.
Update the index on the same day as the performance review, while the comments still make sense. Waiting until “later” is how libraries become junk drawers.
Default private. Share a PDF of the report and a cleaned deck if data is public. Do not share teammate chat. Do not share model solutions. Do not share a certificate for unfinished modules. If a university portal wants a file, pick the report. It travels better than slides.
Keep the editable sheet and the PDF. Keep the deck and a PDF export. If a tool disappears, you still have a report. Students who keep only a link in a school account they will lose in June learn this the hard way. Export on the same day as the performance review.
Name files so a stranger could guess the contents: 2026-04-12-recommendation-v3.pdf, not finalFINAL2. Future-you is a stranger. So is a counselor helping with an application. Evidence libraries fail on naming as often as they fail on missing work.
Once a year, or before application season, delete true junk: screenshots of webinars, duplicate v1s, other people’s model solutions. A library that cannot be scanned in five minutes is a pile. Piles do not get opened. Unopened libraries do not help interviews.
Continue
Each page covers a different part of the remote internship pathway so families can compare structure, work, and evidence.