Company
Whatnot
8 cases from Whatnot (Marketplace, E-commerce, Social), each digested from a public talk or engineering post and linked to the original. 3 were checked in an online A/B test; 2 report results in production.
Tasks: Build a data platform, Recommend to new users and new items, Retrieve and rank from millions of items, Balance relevance and diversity, Rank search results and 2 more. Methods: A/B test, Rule-based heuristics, Gradient boosting, CDC / incremental processing, Pointwise ranking, Feature store and 15 more. Tools: Elasticsearch, dbt, OpenAI API, LangChain, Apache Kafka, UMAP and 3 more.
Tech blog: medium.com/whatnot-engineering
| The New User Dilemma: Why We Killed Our Heuristics (& What We Built Instead) | marketplace | 2025-09 | |
| Evolving Feed Ranking at Whatnot | marketplace | 2025-04 | |
| Whatamix: Blendable feed construction | marketplace | 2024-11 | |
| Powering real-time magic moments through notifications | marketplace | 2024-04 | |
| Enhancing Search Using Large Language Models | marketplace | 2023-09 | |
| How Whatnot Utilizes Generative AI to Enhance Trust and Safety | marketplace | 2023-09 | |
| Feeds with Real-time Signals (Home Feed, Part 2) | marketplace | 2022-06 | |
| Going from 0 to 1 modeling User Preferences for Personalized Recommendations | marketplace | 2022-03 |