No. Sorry, you're not forming a coherent vector space between both skills and demands without published research on the subject.

You do not simply say, "Gosh it seems like I have said words related to ML and therefore its application is plausible."

What's more, the notion that recruiting is in fact a skills demand model is itself fundmentally misleading. Many of the skills you want are domain specific and in fact cannot be expected to be acquired anywhere but on the job. Given how many shops subtly permute "react" or "Golang" to mean a lot of skills in a utility cluster, any space you form is going to be incredibly specific to the employer and difficult to map anywhere else.

>> subtly permute "react" or "Golang" to mean a lot of skills in a utility cluster

This, and the tendency for shops to evaluate candidates' skillset against the shops' most mission critical parts and processes which for whatever reason have not been operationally hardened and locked down.

How does embedding apply here? What's the justification?

It works very nicely for predicting what movie you'd like to watch, so it could potentially work with jobs, too. Collecting enough data could be challenging, though

That's an incredibe over-simplification of the utility and usage of embedding. For an embedding to work there must be some legitimate (as opposed to arbitrary or even non-existent) relationship to be teased out.

> For an embedding to work there must be some legitimate relationship to be teased out.

There is most definitely a relationship between candidate skills, job requirements, and interview result/job performance.

The point is to get rid of all these bullshit subjective excuses when people fail an interview/get fired. The answer is simple: they likely weren't good enough. But no one likes hearing that.

> There is most definitely a relationship between candidate skills, job requirements, and interview result/job performance.

This is actually not at all an obvious fact. It's continuously offered as a ground truth, but many people dispute it and a lot of successful organizations do not recruit weighing these factors as heavily as you're suggesting.

> The answer is simple: they likely weren't good enough. But no one likes hearing that.

Possibly, or possibly they were plenty good but so obnoxious or outrageous that they wouldn't be welcome. I've certainly done that more than once in my time building tech organizations. I still remember the guy who effusively praised the beauty of all the women he saw and congratulated me on "the haul". Even touched a woman's hair to compliment it. Too bad he was such a sleazebag, he seemed smart. But even from a cold economic standpoint the cost to the company for an inevitable sexual harassment lawsuit would always eclipse any value he could provide.

>> a vector of your skills vs. a vector of job requirements

Of course it's an over-simplification, but you don't think there's a legitimate relationship between the two?

Maybe, maybe not. It isn't obvious in any case, and reducing the problem to an embedding in the way suggested requires more validation.