If, at 3:24, your opponent had traded rooks as you mention, it looks like he can play a4 and I don't see how Black can break the bind. There was another spot where you showed a variation where he was able to trade his a-pawn for your b-pawn with your rook in front of your a-pawn which would allow him to get behind this pawn and prevent the promotion. Of course, it is a bit tricky, but not complicated, and I think Black should draw.

Nevertheless, it is a very good presentation of the pitfalls of the Q+R+5P vs Q+R+4P and I urge others to watch this video.

Endgames are the cause of a lot of lower rated player losses. I don't know White's rating in this game but you seemed to be higher rated, so this is a case in point, he had some drawing chances. When looking to improve in chess, equal attention should probably be applied to all phases of the game - and the endgame is part of the game (maybe the most important!).

Yeah, the variation where he traded rooks was a good try for white. During the game, my plan was to bring my queen to the b file, king to f6 or e6, and then play b5. Obviously it's not that easy to execute without allowing checks, but some way for me to torture him forever.

In the R+4 and R+3 that you mentioned (at 8:30 in the video I believe), with the standard pawn structure (white pawns on g3 and h4), white loses unless he can get his king to f3 before the black king gets to e5. Since white didn't have the standard pawn structure and his king would be on the h file, it's an easy win for black. The win itself is really complicated (check out Dvoretsky!) but maybe I'll do a video on that...the basic idea (I didn't want to get too in depth in the video) is to rush the king to b4 when the pawn is on a3, then win white's rook for the pawn with a maneuver like Rc2-c5-a5.

My opponent was higher rated than me in this game (I think ~1900 at the time?) but he's really young and I don't think he had worked on the endgame too much. He definitely could have swapped a pair of pieces to make his life easier.

I felt there many mistakes in your analysis. Like not sure b6 was the best move because he can play Qe7. Maybe Rd7 instead. Also what does Kg7 do? Seems a bit slow and not sure it improves anything. Also the rook ending is not winning with just the a-pawn.