Turbo-Scaling GenAI at DoorDash: From Product Knowledge Graph to Real-Time Personalization
Go behind the scenes with DoorDash’s New Verticals AI team as they share how they’re turbo-scaling generative AI to powering everything from ultra-relevant search and personalization to automated dasher support for their new business verticals such as grocery, convenience, alcohol, and retail delivery.
In this talk, the New Verticals team reveals the architecture that powers DoorDash’s ever-growing Product Knowledge Graph and how fine-tuned LLMs, retrieval-augmented generation (RAG), and agentic workflows turn 100s of millions of product SKUs into precise, real-time recommendations and personalized shopping experiences. They also cover strategies for mitigating inference costs like model cascading and what’s next for context-aware AI that can build your cart before a storm hits. If you’re scaling GenAI in e-commerce, grocery, or logistics, this talk is packed with hard-won lessons you can put to work today.
-------------------------------------------------------------------------------------------------------------------------------------
Download the Slides: https://pbase.ai/doordash_deploycon_slides
Try Predibase for Free: https://predibase.com/free-trial
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
Related AI Lessons
⚡
⚡
⚡
⚡
I Stopped Using Claude Like a Chatbot
Medium · AI
The Future Team May Not Have Dedicated Leads Anymore
Medium · AI
The Linter Has a New Job: Keeping AI Agents Honest
Medium · Programming
AI Agents Can Be Readily Turned Into ‘Useful Idiots’ And Unwittingly Perform Devilish Acts
Forbes Innovation
🎓
Tutor Explanation
DeepCamp AI