Dynamically corrected gates (DCGs) provide a general perturbative framework for boosting the fidelity of gating operations in a large class of noisy qubit settings. While original DCG constructions rely on fully analytical control design and aim to maximize portability with respect to the underlying error model, realistic applications may call for DCG sequences that are both more efficient and more flexible in accommodating control constraints. After reviewing the standard DCG framework, I will show by example how numerical optimization methods can be successfully applied to this end. In particular, I will address a practically motivated scenario where the presence of an always-on drift Hamiltonian mandates numeric search in order to allow robust gate synthesis.