Does your GitHub sell you?

When you type your username, your browser asks GitHub for your public profile and your own repositories (not forks or archived ones, unless you ask). You choose up to six projects. For each one we read the list of files, the README and up to two Jupyter notebooks under 2 MB. Then we score what we found against the rubric below.

Why these things? Hiring managers for junior data roles rarely have time to read a GitHub profile closely. They skim: they open your top projects and read the start of the README. They’re looking for the question you set out to answer, the data, what you did, what you found (ideally with numbers) and whether the work looks recent and your own. Code matters, but a clear README is what gets it read.

The score is out of 100:

• Skills shown (25 points). SQL (10): .sql files or a dbt project. Python or R (8): .py, .ipynb or R files. BI (7): Power BI (.pbix, .pbip) or Tableau (.twbx) files, or Power BI or Tableau named in a README. SQL or Python that’s only mentioned in a README gets half.
• README quality (35 points). For each project: length (7, full marks at 120 words or more, not counting code), the business question or goal (6), the data (5), the approach (4), the findings (3) with numbers (3), how to run it (3), and a chart, screenshot or live dashboard link (4). We average across your projects; well-known practice projects count half.
• Recent activity (15 points). Your newest push: within 90 days (10), 180 days (6) or a year (3). Plus up to 5 for the share of projects updated in the last 90 days (full marks at half or more).
• Notebook readability (15 points). Headings, explanation in markdown between the code, and saved outputs, so charts and tables show without anyone running the notebook. If none of your projects has a notebook we can read, this part follows your README score.
• Presentation (10 points). A bio (2), a profile README, the repo named after you (3), one-line repo descriptions (3), and how many of your projects are substantial data work (2).

A score of 70 or more means your GitHub shows roughly what junior data roles usually look for. The table in your result shows every check for every project, so you can see exactly where the points went.

We flag practice projects gently. Projects that look like well-known tutorials (Titanic, Iris, house prices, Boston housing, Hello World, course exercises) are fine for learning, but they count half towards your README score and aren’t counted as substantial projects.

What it can’t do. These are simple rules, not a person. We can’t judge whether your analysis is correct or your code is good, and a README can tick our boxes without being clear. We can’t see which repos you’ve pinned, your contribution graph or private work. READMEs written in languages other than English may score lower than they should.

How it stays light on GitHub. A check uses at most eight requests to GitHub’s API: your profile, your repo list and one file list per project. READMEs and notebooks come from GitHub’s raw file server, which doesn’t count towards the hourly limit.