How to Organize Your Humanitarians AI Fellows Folder

TL;DR 0:17 Why it matters 0:35 The real question 0:54 Work nobody can find 1:08 Start at the person 1:27 Name folders by date and topic 1:43 Why one giant log fails 1:58 The README is the entrance 2:15 A record someone can follow 2:41 Back to the same fellow 3:00 Verdict 3:16 Your turn

[0:00]Hello, this is Liam in for Bear. This film shows how to organize a humanitarian's AI fellows folder. Keep the first name and last initial naming convention. Open with a clear read me about your research and group. Give every work folder a descriptive name, a report, and its own frictional log. Make the evidence easy to open. Why should you care? Your reviewer sees the folder before hearing your explanation. Vague names and scattered logs make real work difficult to evaluate. Professor Bear is busy and will not hunt for evidence. Your job is to organize the record so another person can follow your contribution, your reasoning, and your actual work. How does my folder make my work easy to verify? The original question focuses on the request. The useful question focuses on what the reviewer can actually verify. Two useful terms, kebab case, lowercase words joined with hyphens, readme.md, the markdown page GitHub opens for a folder. Here's a report, a video plan, and a reflection.

They sit in folders called final, new, and stuff. Which one explains the fellows' latest result? Nothing in those names answers the question. Let's keep the work fixed and change the path a reviewer follows to reach it. Start at the person. The repository uses first name and last initial, all lowercase joined by hyphens. Example fellow is fictional and its demonstration folder is example F. Keep your existing identifier. When names collide, the repository permits an extra surname letter. A public path does not need your full name or private documents. Now keep the fellow fixed and change the work. A dated explainer tells you when and what. A dated progress folder tells you something different, the week's report. Year, month, day keeps dates sortable. Add the work type and a specific topic. The name should let a reviewer predict what is inside. You might put every reflection in one giant log at the top, but then which entry explains the speech sheet? The reader has to match

Dates and guess. Keep the frictional log beside the piece of work it describes. The fellow readme can link to it without duplicating or moving the record. Make the readme the entrance. It states your role, project, group, [clears throat] research question, completed results, and next steps. Beneath it, descriptive work folders branch to their own reports and evidence. Predict where a reviewer should find what went wrong while making one particular explainer. Follow the work, not a separate pile of notes. Watch the log move beside the artifact it explains, and the readme connect directly to both. Now the folder has a record someone can follow. The log says what you tried and expected, where it resisted, and what you did next. It names AI or human help, and what you accepted, changed, or rejected. Finish with what you learned and what remains unclear. Date entries as you work. Label late entries honestly. Return to the same fellow. Now the readme leads to completed work, individual weekly

Hours, and the markdown renewal request. Each folder explains its contents. The example has copied source sheets and missing finished videos. Those limits stay visible. Real fellows replace fiction with actual records. Why is this easier? The author has already done the organizing. Let's recap with Claude. Keep the existing fellow identifier. Name work folders by date, type, and topic. Put each frictional log beside its work. Readme links straight to evidence. Professor Bear is very busy. The fellow writes the clear report. The reviewer should not have to hunt. Return. Paste this into Claude. Inspect my fellow's folder and readme. Can a reviewer identify my research group, contribution, weekly hours, and renewal request? Flag vague folder names and missing links. Suggest an organization plan without moving files, inventing evidence, or rewriting earlier log entries. Then open every link yourself. The useful answer identifies gaps. It does not manufacture a convincing history. Keep your own judgment. What How to organize your humanitarian's AI fellows folder at Nick Bear Brown.