Using Leetcode as a filter is great if you're looking for someone who interviews often, but in my experience it's not a great gauge if someone is a good engineer or not.
As a hiring manager we've tried implementing different strategies. On my team, we hired 4-5 engineers doing live coding exercises similar to leetcode, and none of them made it more than 6 months. We switched to a take-home project for the last 3 we've hired and they've all been with us for 2+ years and have been great.
Of course with the take home projects you're filtering for people who are willing to take a few hours of personal time to work on something, but over the interview process it seems like it washes out? Of course this is just anecdotal data on a fairly small team, but I hate LeetCode style interviews.