During February 2025, this developer built an automated language model evaluation pipeline for the kilian-group/phantom-wiki repository. They designed and implemented cot_rag.sh, a shell script that streamlines end-to-end evaluation of language models by dynamically selecting models based on a user-specified size parameter. The solution automated the startup and shutdown of vLLM servers for each evaluation run, ensuring efficient resource allocation and process isolation. Leveraging Bash and Python, the developer focused on improving automation, reproducibility, and throughput for model evaluation tasks. The work demonstrated depth in DevOps, LLM deployment, and shell scripting, addressing practical challenges in scalable model assessment.