Very fun read! My first try at implementing DLX was for sudoku, as well. I keep meaning to add something showing why it backtracked at a particular spot. Is fun watching the numbers it tries along the way.

And if you haven't seen it, Knuth has a whole section on puzzles in his newer writings. Always fun to flip through.

(If you want to see the numbers fill up, I have that at https://taeric.github.io/Sudoku.html)

Well done! I used to play Sudoko daily, but have switched to KENKEN, and I think a similar approach would work well.