Your Ranking Model Is Optimizing the Wrong Thing with Zoe Weil - #780
Sequen uses live behavioral signals like scroll speed to optimize retail rankings for purchases, not clicks
Sequen co-founder Zoe Weil describes an ML system that infers shopper intent from in-session behavioral signals—scroll speed, hover duration—to rank products by purchase likelihood rather than click-through rate. The architecture uses large event models feeding specialized downstream rankers, with Rust handling millisecond-level serving. While technically interesting, this is a niche retail ML engineering story with no major AI industry signal.