Zhen Wang

Serving deep learning based recommendation models (DLRM) at scale is challenging. Existing approaches rely on dedicated ANN indexing and filtering services on CPUs, suffering from non-negligible costs and missing co-design opportunities. Such inefficiency makes them difficult to support complex m...

Traditional recommendation systems suffer from multi-stage objective misalignment. Generative recommendation offers end-to-end training, yet existing methods are largely inductive and miss complex intents that require world-knowledge-driven deductive reasoning. LLMs reason well but are impractica...