Define a field with 256 elements

Python 2, 11 + 45 = 56 bytes

Addition (11 bytes):

int.__xor__

Multiplication (45 bytes):

m=lambda x,y:y and m(x*2^x/128*283,y/2)^y%2*x

Takes input numbers in the range [0 ... 255]. Addition is just bitwise XOR, multiplication is multiplication of polynomials with coefficients in GF2 with Russian peasant.

And for checking:

a=int.__xor__
m=lambda x,y:y and m(x*2^x/128*283,y/2)^y%2*x

for x in range(256):
    assert a(0,x) == a(x,0) == x
    assert m(1,x) == m(x,1) == x

    assert any(a(x,y) == 0 for y in range(256))

    if x != 0:
        assert any(m(x,y) == 1 for y in range(256))

    for y in range(256):
        assert 0 <= a(x,y) < 256
        assert 0 <= m(x,y) < 256
        assert a(x,y) == a(y,x)
        assert m(x,y) == m(y,x)

        for z in range(256):
            assert a(a(x,y),z) == a(x,a(y,z))
            assert m(m(x,y),z) == m(x,m(y,z))
            assert m(x,a(y,z)) == a(m(x,y), m(x,z))

JavaScript (ES6), 10 + 49 = 59 bytes

a=(x,y)=>x^y
m=(x,y,p=0)=>x?m(x>>1,2*y^283*(y>>7),p^y*(x&1)):p

Domain is 0 ... 255. Source.

Intel x86-64 + AVX-512 + GFNI, 11 bytes

add:
    C5 F0 57 C0     # vxorps     xmm0, xmm1, xmm0
    C3              # ret
mul:
    C4 E2 79 CF C1  # vgf2p8mulb xmm0, xmm0, xmm1
    C3              # ret

Uses the new GF2P8MULB instruction on Ice Lake CPUs.

The instruction multiplies elements in the finite field GF(2⁸), operating on a byte (field element) in the first source operand and the corresponding byte in a second source operand. The field GF(2⁸) is represented in polynomial representation with the reduction polynomial x⁸ + x⁴ + x³ + x + 1.