The potential of genomic selection (GS) to increase the efficiency of breeding programs has been clearly demonstrated; however, the implementation of GS in rice (Oryza sativa L.) breeding programs has been limited. In recent years, we have begun to work towards implementing GS into the LSU AgCenter rice breeding program. One of the first steps for successful GS implementation is to establish a suitable marker set for the target germplasm and a reliable, cost-effective genotyping platform capable of providing informative marker data with an adequate turnaround time. In this study, we develop an optimized a marker set, the LSU500, for application of routine GS in Southern U.S. rice germplasm. The utility of the LSU500 was demonstrated using four years of breeding data across 8,473 experimental lines and four elite bi-parental populations. The predictive ability of GS ranged from 0.13 to 0.78 for key traits across different market classes and yield trials. Comparisons between phenotypic selection and GS within bi-parental populations using the LSU500 provided evidence of the potential of GS to improve the efficiency of a rice breeding program. The design of this marker set followed a continuous integration strategy, whereby GS is initially introduced into a breeding program while technical and strategic aspects of GS implementation are evaluated, optimized, and integrated into the breeding pipeline on-the-go. The LSU500 marker set has been established through the genotyping service provider Agriplex Genomics, and in the future, it will undergo improvements to reduce the cost and increase the accuracy of GS.