In startups, there is a saying that you only need to be right once. You only need one investor to say yes regardless of how many attempts it takes you. During this interview cycle, I was shooting for the stars and looking to get an offer from either Google or Facebook. I interviewed with a total of 7 companies but only struck with Facebook and received an E4 offer today. Since leetcode has given so much to me in the past few months, I wanted to do my part and give back. I had solved 317 leetcode problems and yet there were some online assessments I failed, some technical phone screens, other onsite follow ups, and none of these mattered at the end, because I only needed to be right once with one company. I say all this both to emphasize the fact that it's possible to catch a perform storm if you work hard and that there is no magic number of leetcode problems you must solve that will make you immune from encountering some hiccups during the interview process. So cast your net wide and far. Nonetheless, disciplined leetcode preparation was indispensible to my success and gave me the confidence to confront any problem head-on. Just in case it helps someone, I'll share how I prepared.
Preparation
I had been working in the startup world for the past 2.5 years and I didn't touch leetcode-style problems at all during this time. So I used @Pooja0406's post as a starting point and solved problems from her must-do easy question list and once I finished those I did her must-do medium list. I can attest to the fact that her questions are carefully selected and cover the entire breath of problem solving patterns that one would need to be comfortable with before a FAANG interview.
After I finished those two lists, I jumped to another list: Blind's 75 leetcode questions which were carefully crafted by a person who solved more than 500 leetcode problems. Once I wrapped this list up, I solved problems from Sean Prashad's leetcode-patterns list of 170 problems in random order, never looking at which category a topic fell into.
After this point I started feeling fairly comfortable across all topics, and I had solved about 250 problems in total. In the subequent weeks, I started participating in weekly contests on Saturdays, and the level of ingenuity required to solve contest problems (mainly problems 2-4) really opened my eyes to other advanced techniques. Alex Wice's post-contest review session, Interview Weekly on Youtube, was a fantastic resource that helped me consolidate my learnings.
At the end of my leetcode journey I solved 317 total questions. I reviewed these 317 many times however and I would solve problems I had solved or struggled with many months ago from scratch again. I kept a master list of all the problems I solved, including the date of my last submission, how many times I reviewed the problem, any comments, how long it took me to solve, and I classified problems as Mastered, In Review, and In Queue. Here's an example:

In my experience, the key to building your problem solving muscles is to accumulate as many "aha" moments as you can that will stick with you. Once you accumulate these moments across a breadth of problem types, you will be able to solve new problems by mapping them into previous problems you have solved before. That's why the Blind 75 list and the Leetcode Patterns list was so essentional in my preparation. In order to generate these "aha" moments you really need to struggle with each problem, by giving it your undivided attention for atleast 30 minutes. If the problem is still interesting to you or is still nagging you, you will be able to crack it after you give it some additional time, the next day, or maybe even in a dream (yes really!). Struggling with the problem by yourself also makes sure you get the most out of the discuss section.
I will end my rant here, but I just want to say mad respect and appreciation to everyone in this community who is putting in the work. Keep at it, it will pay off, and remember you only need to be right once!
Update
Since some of you guys asked, here is a link to the list from the screenshot above.