How would you adapt your previous research to run efficiently on an embedded system or a mobile processor at Qualcomm?
