Real-Time Translation System (Text & Voice)

Let me walk you through how I'd design a real-time translation system at production scale handling both text and voice inputs. I'll structure this to make it easy to pivot between text-focused and voice-focused discussions, since the core translation engine is shared but the preprocessing and serving constraints differ significantly. I'll go stage by stage – business and ML objectives, high-level design, data and feature strategy, model architecture, infrastructure, evaluation, and robustness.