The state of each qbit is represented by a state vector of two complex numbers [a, b] where |a|^2 + |b|^2 = 1. There are two special qbit values called the classical basis: [1, 0] which is the classical bit 0, and [0, 1] which is the classical bit 1. If a qbit is not in one of the two classical states, we say it is in superposition. When a qbit is in superposition, we can measure it[0] and it will collapse probabilistically to 0 or 1; for a qbit [a, b], the probability that it collapses to 0 is |a|^2 and the probability that it collapses to 1 is |b|^2.
Things get more interesting when we have multiple qbits. If we have two qbits [a, b] and [c, d], we define their product state as their tensor product [ac, ad, bc, bd]. For example, if we have two qbits both in state [1/sqrt(2), 1/sqrt(2)], their product state would be [1/2, 1/2, 1/2, 1/2]. We use the product state to calculate the action of a quantum logic gate that operates on multiple qbits - for a gate which operates on two qbits, we can always represent its action as a 4x4 matrix.
Usually we can move back and forth between the product state representation and writing out the individual qbits states. However, in certain scenarios something very special happens: we cannot factor the product state back into the individual state representation! Consider the product state [1/sqrt(2), 0, 0, 1/sqrt(2)]. If you try to write this as a tensor product of two states [a, b] and [c, d], you cannot! It cannot be factored; the qbits have no individual value, and we say they are entangled.
Well, what does this mean? It means when you measure one qbit, even if the qbits are very far apart, you instantly know the value of the other qbit. So if I entangled two qbits in the state [1/sqrt(2), 0, 0, 1/sqrt(2)], give you one, then we go to opposite ends of the universe, if I measure my qbit and see a 0 I'll know your qbit instantly also collapsed to 0 (or collapsed to 1 if I measured 1). This phenomenon has been experimentally-verified to occur faster than light. It is instantaneous, as far as we can tell. So, local realism is wrong! Spooky action at a distance is real!
There is an important caveat: while the qbits seem to coordinate in some faster-than-light way, you cannot use this to communicate in a faster-than-light way. All we have is a shared random number generator. I can't send some chosen bit from my reference frame to yours. This is called the no-communication theorem.
If you found this interesting, I have a full video on quantum computing for computer scientists here: https://youtu.be/F_Riqjdh2oM
[0] IDGAF about your chosen quantum mechanics interpretation, don't @ me