Nowadays, anyone can train a neural network.
Yes seriously. Even with no-code solutions and get very passable results. I've written about imposter syndrome before and how I think it disproportionately impacts machine learning people. But there's one skill that sets you apart from other folks doing machine learning.
Learn how to communicate with stakeholders.
I talked my way into a job
My current job is quite specific in applying machine learning to weather modelling.
I had no idea about weather modelling during the interview. I was extremely transparent about this fact. In the end, I got the job for my communication skills. Anyone that applied was better in numerical weather prediction. Some were better in their theoretical understanding of machine learning. Others were more senior and had more business acumen.
Few, however, understand how important proper communication of machine learning results is.
No one cares about your model or accuracy
I get it.
Finally building a working model with a few tricks I just learned from the latest papers is exhilarating. Stakeholders and colleagues mostly don't care. Most people want to know the impact of your model. How does it perform on the problems that are hard for the person you're reporting to, to eventually make their life better?
The expectation of your colleagues and stakeholders is for machine learning to take over hard tasks. It's to make tangible improvements to their part of a business or research question.
This can be disheartening if you expect to nerd out about your ML solution.
What to do instead
I'm currently preparing a full talk for Pydata Global 2021 (in case you read this later).
It's a cheeky take on the issue called "How to Guarantee No One Understands What You Did in Your Machine Learning Project". Where I explore specific tools from machine learning explainability, over to ethical considerations and visualizations, all the way to model validation specific to different expert domains and tie it all together.
It's actually a lot of fun once you wrap your head around it!