5th edition assumes about four players. Two works, and often works better — more spotlight each, easier scheduling, a story that can actually be about your characters. But three things break at that size, and all three are fixable without rewriting the rules.
Do not simply cut monster hit points — that makes fights shorter without making them safer. Cut the number of enemies and raise their quality. One dangerous opponent and two weak ones plays far better at a small table than five identical bandits.
Other things that work:
Three approaches, in the order most tables should try them:
Whichever you choose, make healing available somewhere. A potion in a pack counts; no answer at all does not.
Small tables are not a compromise. Two players means every session can be about those two characters — their debts, their history, the people they left behind. There is no waiting through three other people's shopping trips, no fighting for airtime, and scheduling two calendars is trivially easier than five.
Duet play in particular produces the kind of campaign people remember: closer to a novel than an ensemble, with consequences that land because there is nobody else to absorb them.
Everything above applies more sharply. Encounters should mostly be one-on-one or one against a couple of weak enemies. Give the player a companion so a single bad roll does not end the campaign. And lean toward investigation, negotiation and exploration, which scale down far more gracefully than pitched battles.
The other small-table problem is that one of your two people has to run the game, and that person never gets to play a character. That is what an Autonomous Dungeon Master solves: both of you play, and the world is run for you.
MONOLITH handles one to six players on a single screen, or joining from their phones with a room code. It tracks initiative, hit points and death saves, scales encounters to the party it actually has, and hands you the dice when a check is called for. The first session is free and solo — 24 rounds, about an evening.