Building ML Pipelines with Open LLMs