I interviewed 23 candidates for an ML engineer role last quarter.

19 of them had almost identical resumes — TensorFlow certification, Coursera ML specialization, a Titanic dataset project on GitHub.

The 4 who stood out? They'd actually built something messy and real.

One had a RAG chatbot for her grandmother's recipe collection. Another built a model that predicts when his houseplants need water (it was hilariously overengineered). The third automated his fantasy football drafts with ML.

None of these were "impressive" by traditional standards. But they showed something certificates never will — curiosity and the ability to ship.

If you're trying to break into AI/ML in 2026, stop optimizing your resume and start building something weird that you actually care about. The projects that get you hired are the ones that make interviewers lean forward and say "wait, tell me more about that."