You are asking for 0.01K resolution rather than accuracy, which seems appropriate. It implies about 0.5 uV voltage
resolution. It's generally feasible, but low frequency noise and offset drift can endanger it. Unfortunately, the AD595 has
no noise specification. If the input stage is similar to other ADI instrumentation amplifiers, it should work. In any case, a
suitable low-pass filtering respectively averaging should be provided.
Other aspects are the temperature range and repeatability. Most of these metal type temperature sensors are sensitive to shock and vibration and there is some hysteresis.