My point is that LeetCode and buzzwords became the unofficial benchmark for a software engineer during the distorted/ZIRP era. You had to be good at it to get in, and once in you could just coast your way through, not that anyone would mind because so many of your peers and superiors are doing the same thing and none of it matters as the company itself may be coasting on investor money and never intending on making a profit.

Unfortunately, LeetCode skills and "architecture astronomy" (https://en.wikipedia.org/wiki/Architecture_astronaut) are a very poor proxy for actually driving profit. This doesn't mean a LeetCoder (for fun or to play the game - as long as they understand it is a game) can't also be a good contributor in today's market, but their value would come more from experience solving actual business problems and ability to collect/define/argue requirements and then translate them into a system (which may or may not even be code - sometimes a Google Form is all that's needed) rather than whether they know LeetCode.

When it comes to solving business problems and actually driving profit, programming in abstract terms (the kind that LeetCode and most interview processes select for) is worthless - LLMs can do it just fine. The hard part is understanding/fleshing out the business requirements (including being aware of what you don't know, and planning for that or seeking outside help), refining your solution with the relevant stakeholders, and being aware of high-level constraints including regulatory requirements, security, etc. Once you have a good enough specification, programming it is the generally the easy part.