A reverse proxy sits in front of clients, receives their requests, forwards them to one of several backend servers, and returns the responses.
Clients see only the single reverse proxy without knowing the real backends, and this vantage point handles load balancing, caching, TLS termination, and compression.
The key to the name is that it's the opposite direction of a forward proxy. A forward proxy acts on behalf of clients going out to the internet (corporate firewalls, user privacy); a reverse proxy acts on behalf of servers, accepting requests from the internet. So a forward proxy hides who the client is talking to from the server, while a reverse proxy hides which backend answered from the client.
In practice, software like Nginx, HAProxy, and Envoy is widely used as a reverse proxy. It terminates TLS to centralize certificate management, caches static content, spreads load across many backends, and applies header manipulation, request routing, and rate limiting. API gateways and CDNs are essentially reverse proxies with extra features layered on. By hiding backend IPs, it also serves as a security boundary.