Developed and delivered a dynamic resource specification feature for SLURM job submissions in the pytorch-labs/monarch repository, enabling users to allocate proportional CPU and memory resources rather than requiring full node assignments. This enhancement, implemented in Python, improved backend job scheduling and resource management by allowing more granular and efficient use of cluster hardware. The work involved updating submission APIs and CLI tools to support flexible resource requests, facilitating increased scalability and reduced hardware waste for diverse HPC workloads. The feature underwent a thorough code review and cross-team collaboration, resulting in a stable, production-ready addition to the monarch platform.