RQ-VAE

Hold

Techniques

A vector-quantization method used to learn semantic IDs for representing catalog items.

Why it's here

Placed in Hold: 1 article(s) of evidence from 1 source(s), led by research-stage coverage, with 0 in the last 30 days. Confidence 24%. Low accumulated evidence, so it defaults conservatively pending more signal.

Evidence (1)

6InfoQ·8/15/2026research
DoorDash details shift from one-shot predictions to context-aware AI agents
Sudeep Das explains how DoorDash is evolving its recommendation system from legacy one-shot predictions to an agentic platform that can use consumer context over time. The approach combines language-native consumer memory, RQ-VAE semantic IDs for catalog representation, and grounded search to improve relevance and conversion.