I've thought that in N2, its symmetric so the expansion coefficients would be equal from the pi-bonding MO
Yes, that's right; symmetric molecules have symmetric MOs which have symmetric coefficients of the constituent AOs.
And for CO, the coefficients could be different ...
And yes, conversely for unsymmetric molecules, the coefficients are going to be different.
... with the oxygen having less character due to it being more electronegative?
This is where it starts to get rather murky.
In this specific case, where the π MOs are formed from carbon and oxygen 2p orbitals only, then you can say that the bonding π MO has more oxygen character (not less!). This is because the oxygen 2p orbitals are lower in energy, so the bonding MO (which is lower in energy than antibonding) will take on more of the oxygen 2p than the carbon 2p. On the other hand, the π* antibonding MO will have more carbon character.
(You could say that this is because of electronegativity, but I think that effective nuclear charge is a slightly better rationalisation. That said, they're all interrelated, of course.)
However, this line of logic only works when you have the case of two AOs combining to form two MOs. The moment you have more than two, you need to be very careful about claiming which one has greater contribution. For example, the σ framework of the same molecule, CO, turns out to be pretty complicated, and it's not obvious at all which combination of carbon 2s, carbon 2p, oxygen 2s, and oxygen 2p forms any given MO.
Also, even in the 2 AOs/2 MOs case, the electronegativity argument must be applied with caution. We could say above that oxygen 2p is lower in energy than carbon 2p, but if you're not comparing the same orbitals, then there's no guarantee that the more electronegative element has orbitals that are lower in energy.
Generally, if you wanted to make definitive statements about which atom had a greater contribution, you'd have to actually use a computer programme to calculate these actual contributions. This is known as electronic structure theory, which is a fairly prominent subfield of computational chemistry. If you don't have something like this, then the conclusions you can draw in a qualitative fashion are fairly limited.