rga uses pdftotext (from poppler) internally for pdfs, except wraps it in parallelization and a very fast cache layer, since you usually want to do multiple queries per file :)