I work in the oil and gas industry, and in mine and other heavy industries, the need of the day seems to be automating previously manual tasks, doing descriptive analytics and using simple models for the most part. For example, in dealing with a lot of time-series sensor data, I have seldom seen LSTM being successful because of the complexity of physical processes involved. Instead, simple fitting models incorporating some physical understanding are what works most often. So, the need seems to be of someone who has enough domain knowledge + good programming skills + good data skills. Yet, job postings often call for experience with complex models including deep learning which may be seldom used in actual work. How to get your foot in the door in these sorts of situations?
Maybe you can apply to the job and in your cover letter, point out why you think those might not be the best approaches and point them to notebooks/blog posts where you solved a real world problem (likely somewhat toy problem though) and how you think that experience makes you perfect for the job.
I guess if you were going to otherwise pass up the application, it couldn’t hurt?