Here are some solves. I got really nervous, made some mistakes, but its ok, so the main problem is the very long pauses.
I already learned a lot watching my solves. I never use the inspection time, just jumping into the solve and looking around the cube all the time.
Plan is to work on cross blindfolded, faster oll recognition from every angle, maybe focusing for auf, faster alg execution might come with time and less parctice, maybe take a day off.
Are two more things im struggling, cant do fast sledgehammer and cant do fast v perm y rotation, takes 1.5s just the regrip after rotation during the alg. Tried the rotationless version, but its really slow.
As you noted, you don't really use inspection time. I'd recommend trying to plan out at least 3 pieces of the cross, and optimally the entire cross in inspection. If you do that, then you should be able to solve the cross very quickly.
For F2L, it seems that you generally have good F2L solutions(I didn't reconstruct them all, but after a couple rewatches they still seem good), but slow execution. You should try turning faster, which you can accomplish by just knowing the F2L solutions better(which comes from practice). Until then, just make sure your solutions are good by doing slow solves and looking at some F2L alg lists(you don't need to memorize the algorithms, but you can compare them to yours and look for new tricks). You also seem to be having a lot of unnecessary AUFs while finding pairs. Try to recognize a case from all angles, that way you can just AUF the the correct angle and do it. You should also be capable of doing a case from both the left and right, as well as back and front if possible(some cases with a bunch of F moves don't work well from the back, since F moves turn into B moves. For those, you'll want to do a single rotation so it's a different solution). If you learned intuitive F2L, then this should be relatively simple to do, as it's just mirrored. If you learned F2L algorithmically however, this could be more difficult. Learning to do this will also greatly reduce rotations, especially double rotations(y2)
From the video, it looks like you know full OLL and PLL, which is a good start! Most people don't learn those until later, so you're definitely ahead of the game. While you do know 2LLL though, it seems that your algorithm speed could be improved. You can drill your algorithms by going to jperm.net, which is a great cubing tool with a good alg trainer. In addition, you should improve the recognition speed. This mostly comes with practice, but you can also try to find patterns on cases that are unique to them that are easy to spot.
V-Perms can be difficult, I'd recommend looking at a few of the main algs. J Perm has a video highlighting each one, which can be found
here.
In short, I'd recommend:
Practicing using inspection to do cross(do it BLD)
Reducing AUFs and rotations in F2L by learning to do cases from the left, right, and back
Drilling OLL & PLL algorithms using jperm.net to improve speed and recognition