Valid Palindrome

This is a two-pointers problem where the key is knowing which move can only improve things. It's a great warm-up that builds invariant-driven scanning without a ton of moving parts. Keep the invariant in your head and the pointer moves become obvious.