Everyone says the market is rough. Layoffs, hundreds of applicants per role, and a pile of CVs that all say "RAG, agents, LLMs".
Here is my honest take on standing out. It is not a growth hack. It is a ladder, and most people stop on the first rung.
1. Claim (everyone does this)
"I know RAG and AI agents." True for a lot of people, and impossible for a reader to check. This alone does not separate you.
2. Show (fewer people do this)
A working repo someone can clone and run in five minutes. I built the AI Engineer Roadmap: 132 topics, runnable Python demos that work offline with no API key. If someone can run it, they do not have to trust me.
3. Prove (very few do this)
This is where the gap is. For my open-source library mcp-safe-actions I did not stop at "it works":
- 101 tests, including real stdio tests that spawn the server
- a mutation check that breaks the code on purpose: 42 deliberate bugs, all caught
- a README section called "what it does NOT guarantee"
Writing down what your work cannot do is the most credible thing you can put in a README. It tells a reader you understand the limits.
What I'm doing in practice
- Build in the open. One real project beats ten tutorials copied from a course.
- Write about it. A post explaining a decision shows how you think, which a CV cannot.
- Take small contributions seriously. When someone opens a PR on your repo, reviewing it properly is part of the proof.
- Stay honest about numbers. I will say "0 reactions" before I invent a success story.
What this does not fix
It is slow. It does not beat a referral. A good portfolio does not guarantee interviews, and I would be lying if I said otherwise. It only makes sure that when someone does look, there is something real to find.
Your turn
What is the one thing you built that you can point to and say "run this"? Drop the link below. I would like to look.

