People You May Know

Let me walk you through how I'd design a "People You May Know" (PYMK) system at production scale. I'll go stage by stage — business and ML objectives, high-level design, data and feature strategy, embeddings, pipeline architecture including modeling and training details, infrastructure, evaluation, and robustness. This is fundamentally a graph edge prediction problem operating on the platform's social graph with billions of nodes and trillions of potential edges.