Notes

What I learned making my own work checkable.

Which projects carry hiring signal, how to prove a system actually works, what to do when the repo has to stay private, and how to run all of it on nothing.

All posts.

04
  1. 9 min

    The AI engineer portfolio projects that actually get you hired

    Two projects you can defend for ten minutes beat eight you can only describe. Here is what separates a portfolio that gets a callback from one that gets skimmed.

    read

  2. 11 min

    RAG with an eval harness: the questions interviewers actually ask

    RAG is the most common project on AI engineering resumes and the one candidates are least able to defend. The gap is always evaluation.

    read

  3. 8 min

    How to make a private repo credible to a recruiter who cannot see the code

    Work you cannot show is not the same as work you cannot prove. The trick is to move the evidence out of the source and into artefacts a reader can check.

    read

  4. 10 min

    Building an ML portfolio on $0: the free-tier stack, and where it breaks

    You do not need a budget to build a portfolio. You do need to know which free tier fails first, because designing around the wrong one wastes weeks.

    read