In this lecture, Zach explains streaming pipeline design, focusing on why Kappa architecture and Delta Live Table have become more practical than older Lambda approaches. He contrasts Lambda streaming plus batch true up for correctness versus Kappa’s single streaming pipeline for low latency, noting Lambda doubles code and was historically needed due to limitations like Hive metastore lacking append. He explains how Python UDFs hurt performance because Python. He also describes Delta Live Table as an orchestration layer like Airflow that defines DAG dependencies, discusses medallion bronze silver gold concepts, and warns about DLT drawbacks such as cost for low-volume data and limited streaming join support.