Jamie Callan
What retrieval-augmented generation looks like when you actually know the thirty-year history of information retrieval it's reinventing.
Callan has been building search engines and retrieval evaluation corpora since before "RAG" existed as a term — Indri, the ClueWeb09 through ClueWeb22 datasets, the kinds of artifacts that NLP papers cite as evaluation infrastructure without realizing they were the original IR research. His more recent contribution as co-author on FLARE (Forward-Looking Active RAG, 2023) is the bridge: an IR foundational figure helping a younger generation of NLP researchers recognize which retrieval problems are genuinely new and which were solved twenty years ago. Reading him is a way to avoid the standard NLP-side mistake of treating the retrieval layer as a black box that "just does search".