A large class of problems in the current era of quantum devices involve interfacing between the quantum and classical system. These include calibration procedures, characterization routines, and variational algorithms. The control in these routines iteratively switches between the classical and the quantum computer. This results in the repeated compilation of the program that runs on the quantum system, scaling directly with the number of circuits and iterations. The repeated compilation results in a significant overhead throughout the routine. In practice, the total runtime of the program (classical compilation plus quantum execution) has an additional cost proportional to the circuit count. At practical scales, this can dominate the round-trip CPU-QPU time, between 5% and 80%, depending on the proportion of quantum execution time. To avoid repeated device-level compilation, we identify that machine code can be parametrized corresponding to pulse/gate parameters which can be dynamically adjusted during execution. Therefore, we develop a device-level partial-compilation (DLPC) technique that reduces compilation overhead to nearly constant, by using cheap remote procedure calls (RPC) from the QPU control software to the CPU. We then demonstrate the performance speedup of this on optimal pulse calibration, system characterization using randomized benchmarking (RB), and variational algorithms. We execute this modified pipeline on real trapped-ion quantum computers and observe significant reductions in compilation time, as much as 2.7x speedup for small-scale VQE problems.