How to produce numba CUDA executable to submit to a cluster job scheduler?

In the traditional C CUDA programming, I compile the .cu code to executable using nvcc and submit it to the SLURM cluster job scheduler using a job script and appropriate #SBATCH directives eg.

#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --partition=gpu
module load cuda
./myexe

However, since python is interpreted no executable is produced. How should such code be submitted to the scheduler?