I'm completely new to the topic, I've read that Ethernet sends Bytes in a Big-Endian way, but that bits are sent with Little-Endian. I've also read that IPv4 uses Big_endian order. So my question is, how can IP protocol stay on an Ethernet layer if they (IP layer and Ethernet) manage bits in different ways?
I wouldn't use the term endianness here. There's no general word size like with CPUs.
Ethernet sends bytes as a bit stream, with the least significant bit first, with the notable exception of the FCS. In multi-octet fields, octets are ordered most significant octet first.
IP uses the same most-significant-octet-first order, where that matters. It doesn't (really) define a bit order as it interfaces with the underlying protocol on the octet/byte level.
For Ethernet, the IP packet with headers is only payload, a bunch of bytes, so whatever rules apply within the payload doesn't matter. The only thing that matters is that the bit and byte orders are well defined and apply to the whole network, to enable data exchange.