When Your LLM Hates Your Data: Building a Self-Healing Airflow Pipeline with Ollama

A deep-dive into designing a batch NLP pipeline that repairs dirty data before inference — and survives when the LLM engine goes dark. Continue reading on Medium »