How would you optimize a signal processing algorithm for execution on an embedded processor in Aeva's sensing pipeline?
