I'm sure it still requires a huge amount of talent and isn't lacking in meaty, difficult problems, but the core product just seems so unimportant.
Actually FB became a haven for xooglers that wanted to escape deteriorating engineering at lower levels at Google; there is still burning spirit of hacking present, even after IPO. FB works hard to retain it (outside noisy open offices though).
If you ever went through engineering interview process at FB, you'd have noticed it's one notch better prepared than the one at Google. Their Glassdoor rating is even slightly better and there are very few companies that are rated better (and usually smaller like Odersky's Lightbend, JetBrains...), so there is almost nowhere to go for a better engineering experience.
From my experience, and based on people I know that work at Facebook, e.g., getting your foot in the door is just a matter of studying medium/hard-level problems on Leetcode (speaking verbatim here). I'm not sure you need a strong foundation in systems programming or design, or even a strong track record of building great software to get hired at these kinds of companies. The metric is mostly Leetcode; it's for efficiency I guess, because of the sheer number of applicants.
I've also worked closely with a few people from such companies. I'd say only one was a great engineer. And he definitely didn't have a high opinion of his former co-workers.
The one company whose engineers I'd probably give the benefit of the doubt would be Netflix, because they're so intentional about "performance".
If someone is a senior engineer at one of those companies, I assume they have a reasonable amount of talent. It's not easy to fake it or avoid expectations as a senior engineer.