Search Ranking

Let me walk you through how I'd design the platform's search ranking system — the ML system that decides which listings appear, and in what order, when a guest searches for accommodation. 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 a two-sided marketplace ranking problem, which fundamentally changes the design compared to traditional web search or single-sided recommendation. You're not just optimizing for the searcher — you're simultaneously optimizing for host satisfaction, marketplace liquidity, and long-term platform health, and these objectives routinely conflict with each other.