Maybe the other knight ought to have gone to f3. Against that, I was looking at 13...Bf5 (threatening ...Ng3 or ...Nxf2, but also with ...Qa5ch in mind, followed by ...Rc8.

At least for me seems that Ndf3 is equally bad.. Black could move Bf5 or even worst, f6! and nobody could have stopped that storm

Nb5 didn't do much. I agree with Nd2 as a better defense.. if it's possible.

The resignation was a bit unexpected, but perfectly reasonable, on reflection. At the time, the lines I played through seemed to win Black a pawn only, but I formed the impression that White's position was left shattered and vulnerable. Looking at it now, I think I must have forgotten that Black wins at least the Exchange as well. After 15...e4 16.Qb5 (say) 16...Bxb2 and now, if 17.Rb1 exf3 and White comes out a piece behind (I suspect I simply miscounted the pieces afterwards in this line, and forgot Black {would have} had the extra bishop). So: 17.Nfd4, but Black can now pick up the Exchange by 17...Bxa1 if he wants (which I probably would have played), or improve his position before doing so by 17...a6! 18.Qc5 Qf6ch 19.Kg3 Bxa1 20.Rxa1 Be6. With only a knight in exchange for rook and pawn, White would have had too big a mountain to climb. Thanks snowman - Tim - for a fine post-game annotation and narrative.