[greyhound_7]

One strategy would be to go after ML/AI jobs that are at the profit centers of companies and whose success or failure means the success or failure of the company. There are lots of AI/ML projects done at the whim of some VP in a random business unit "because we have so much data there must be something there", or working on a speculative product. Those types of projects are really interesting and can be a lot of fun, but they get cut pretty fast when it's time to trim the budget.

I don't have a PhD in AI/ML, but I have delivered ML models into production. Doing so taught me that you must go about your modeling and science work with a practical urgency for results compared to the pace of academic research. Business applications don't require you to prove how smart you are (people just assume it), but they do require compromise to meet requirements and exceed stakeholder expectations. Where a lot of ML people lose the plot (and financial reward) is that they don't make enough tradeoffs for the operational or end user concerns for the model in its entire relevant context. I've seen this manifest a few ways, but a common one is getting fixated on the data you have, but never realizing you need to "close the loop" and make something actually useful for an end user in a measurable way.

Those are the practical, "productionize AI" jobs and there is huge interest in those now given the huge interest in LLMs. Who cares about the downturn, LLM start ups are the hot thing.

There are also industry research jobs, definitely worth applying for, but from my understanding they are very difficult to get.