At Los Alamos National Laboratory, I design and operate a bare-metal GPU cluster serving LLM inference to researchers. I implemented a self-hosted LLM platform using vLLM, Ollama, Milvus, OpenWebUI, and LiteLLM.
I also administer production Linux servers and scientific-computing infrastructure, including containerized services and redundant storage. I built an infrastructure-as-code service for reprovisioning immutable Linux hosts, and streamlined AWS resources to cut monthly costs by over 30%.

