ecommerce_ops · ecommerce · workflow
DoorDash builds LLM-powered consumer, merchant, and item profiles to enable explainable personalization at scale
DoorDash's entity representations relied on opaque embedding vectors that were impossible to explain to consumers or debug, limiting the platform's ability to build interpretable personalization and recommendation features.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Input data gathering
Order history, menu metadata, store ratings, images, and other signals are gathered, cleaned, and structured for each entity type.
Tools used
LLMsPySparkFabricatorAmazon S3PortkeySQLSpark
Outcome
LLM-generated narrative profiles enable explainable recommendations, editable consumer preferences, and batch processing at scale, cutting per-profile spend by roughly 50% and processing millions of records within an hour.
Results
Time savedfraction of the time it would take to do so manually
Cost replacedroughly 50%
Grounding & classification
Source type: technical build writeup
30 fields verified against source quotes.
content generationpersonalizationrecommendation systemsummarizationknowledge baseproduct catalogbuilder submittedhuman review describedmetric backednamed customerproduction runtime claimedtools describedworkflow describedecommercecost reductionemployee productivitythroughput increasetechnical build writeupecommerce opsmarketing opsdata sync enrichment