I have a 3d accelerometer and a 3d gryoscope (measure angular
velocity) and I want to use a kalman filter to figure out the
orientation (xz, yz angles) of the board. 

We can safely assume that the board is not moving. The accelerometer
only has noise, the gyroscope has noise and (a drifting) offset.
Not sure if this is right ..

The standard deviation of the sensors noise is given. 

1) First we need to find out how to convert accelerometer to angles
2) We need to relate these angles to the angular velocity of the
gyroscope. The frame of reference of the two may be different
3) Then we deal with the offset
4) What are my states? Setup the (state space and measurements) equations

Probably this is not going to be linear.

I can answer this question very well if you put it back.