Recruiters and mentors spend about thirty seconds on a GitHub profile before deciding whether to spend five minutes. Most student profiles waste those seconds on a wall of tutorial repositories. Mine has 24 public repositories; three of them do the work, and the profile is arranged so those three are what you see.
Pin three, not six
One retrieval project, one fine-tuning or vision project, one with real users. Pinned repositories are the whole first impression; six dilute it. Each pinned repository needs a description line that states the problem, not the stack.
The profile README
- One line of positioning and where you are based.
- Three projects with a one-line outcome each and a link — the same three that are pinned.
- Wins and sessions with institutes and years.
- How to reach you. An email, not a contact form.
Project READMEs that survive ninety seconds
- First line: the problem and who has it.
- A screenshot or a short recording of it running.
- The numbers: users, evaluation results, performance.
- How it works, in one diagram or five bullets.
- How to run it, and a link to where it is running.
- What you would do next, honestly.
Archive the rest
Tutorial clones, abandoned experiments and forks you never touched can be archived or made private. A profile with five good repositories reads as focused; one with forty half-finished ones reads as noise, however hard you worked on each.
Commit history is read too
Regular, descriptive commits over months signal someone who ships. A single "final version" commit signals a copy. Push as you build — it is also the habit that saves you when a laptop dies the night before a demo.
Fix the profile in an evening, then keep it current with each project from the roadmap. It is the first thing I open when a student asks me for feedback, and the first thing a recruiter opens when they consider an interview.
— Pranjul Rathour, GenAI Engineer from Kanpur, India. Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at any campus: pranjulrathour41@gmail.com.
