RX and TX cause all kinds of trouble because the naming is not totally consistent. In some cases, devices which are intended to be peripherals to a host will use host-centric naming. This means that the peripheral device will have an output called RX which is intended to connect to MCU RX input. And likewise it will have an input called TX which is intended to connect to the MCU TX output. Very confusing!
As a general practice, I recommend you double-check whether the TX and RX pins are inputs or outputs and make sure you connect input on one chip to output on the other and vice-versa.
In this specific case, the pin description table clearly shows that the CP2102N TXD pin is an output, and RXD is an input. So this means that they must be crossed over to connect to your MCU.
Historically, in the good old days when computer programs were written on punchcards, IBM categorized devices participating in UART serial links as "Data Terminal Equipment" (DTE) or "Data Circuit-terminating Equipment" (DCE). The nomenclature for the data lines was DTE-centric. So TxD was an output from the DTE and an input to the DCE. And conversely for RxD. The classic example of a DTE is a terminal on a mainframe, or a PC that is connected to a modem. The modem is the classic example of a DCE.
But when you connect your PC to a microcontroller to debug it, they are both DTE's. So crossover is required.
I believe that cellular modems still use the old-fashioned conventions. The cellular modem is a DCE, so its TXD signal is an input and its RXD signal is an output.
Probably more than you ever wanted to know about that.