1. Do not waste too much time trying to obtain certifications. You might end up spending quite some cash for no real added value (at least interview-wise)
  2. A ML engineer’s job is useful if and only if other people can interact with it. Always think of what comes before and after your model is good enough.
  3. Python is your best friend, but not your only friend: other languages and frameworks might serve better purposes (e.g. create interface with HTML/CSS/JavaScript, containerize your app with Docker, run fast code with C or Rust, etc.)
  4. Get your hands very dirty: it’s the only way to learn ML
  5. Pick a problem you like, “solve it” and document everything on github. It looks much better than a certificate
  6. Even if working alone, treat your project as a team effort: document everything, create branches when modifying your code, ensure reproducibility. Remember that to excel at ML you also need to be a good software engineer. Keep this in mind and You will be better than 99% of candidates applying for roles.
Gemmo's noise classification case study with Sonitus