The author says he is in AI, and when I was hiring entry level ML engineers I had similar challenges. As much as we blame the recruiting pipeline, I think it is the educational pipeline that is not creating a sufficiently diverse talent pool. Part of it is "weed out" courses that adversely affect students from less privileged backgrounds[1]. An additional (perhaps controversial) opinion I have is that companies are so aggressive about their individual diversity goals that they often pluck students out of the training pipelines prematurely (e.g. courting Ph.D. students before they finish their dissertation).

[1] https://www.brookings.edu/blog/brown-center-chalkboard/2021/...