Over the past school year and summer at UMBC, ME S-STEM Scholar Turibius Rozario (he/him) conducted research on gradient descent and gradient-free optimization techniques for neural networks under the mentorship of Dr. Ankit Goel. He implemented standard gradient descent using both pre-existing packages such as TensorFlow as well as his own scripts through MATLAB. Additionally, for simple problems, he demonstrated that solving a system of non-linear equations is a technique capable of exceeding gradient descent in optimization rate and quality. He identified random search method, though primitive, to be able to successfully optimize multi-layered problems with no slope computations.