Inside Perplexity's RAG Pipeline -- How 38 Engineers Built a Web-Scale Search Engine

This module is a deep dive into Perplexity's technical architecture -- the five-stage RAG pipeline, the partnership with Vespa for indexing 200 billion URLs, the model-agnostic orchestration layer, and the ROSE inference stack. What makes this remarkable is that a team of roughly 38 engineers built a system that competes with Google's search quality. Understanding these engineering decisions is essential for system design interviews and for demonstrating technical depth in behavioral rounds.