np.array([ BigInt list ], dtype=object)

then you can apply this bit parallelism to a tensor.