a directed acyclic graph (DAG) of operations. The computation only happens when you call `.compute()`. This allows Dask to optimize the execution plan: it can fuse operations, avoid unnecessary intermediate DataFrames, and parallelize independent tasks.