Spartha developed the ModelService Helm Chart Deployment Proposal for the llm-d/llm-d repository, focusing on reproducible and scalable deployment of large language models within Kubernetes environments. The work introduced a declarative approach to managing both base models and LoRA adapters, integrating seamlessly with the existing llm-d ecosystem. Spartha’s proposal detailed motivations, goals, and implementation strategies, providing clear documentation in Markdown to guide future development and adoption. By leveraging Helm Charts and Kubernetes, the solution addressed the need for consistent, maintainable LLM deployments. The depth of the proposal demonstrated a strong understanding of infrastructure-as-code and modern deployment practices for machine learning systems.