How Pinterest Uses Embedding

The article reflects on a decade of AI platform development at Pinterest, detailing the evolution from fragmented machine learning stacks to a unified AI platform that supports various models.

The article discusses the implementation of Offline Approximate Nearest Neighbors (ANN) at Pinterest to improve ad retrieval efficiency.

The article discusses advancements in embedding-based retrieval at Pinterest's Homefeed, focusing on improvements such as feature crossing, ID embeddings, and serving corpus upgrades.

The article discusses enhancements made to Adaptive Bitrate (ABR) video performance at Pinterest, focusing on reducing startup latency and improving user engagement through innovative solutions.

The article discusses the development of Pinterest Canvas, a text-to-image foundation model aimed at enhancing existing images and products on the Pinterest platform.

The article discusses Pinterest's commitment to inclusivity through its technical advancements in machine learning, particularly focusing on body type representation alongside existing skin tone si...

The article discusses the implementation of streaming filters in the Manas search engine at Pinterest, which utilizes Hierarchical Navigable Small World graphs (HNSW) for Approximate Nearest Neighb...

The article discusses SearchSage, a search query representation model developed by Pinterest to enhance the relevance of search results and user engagement.

The article discusses SearchSage, a search query representation developed at Pinterest to enhance search retrieval and ranking.

The article discusses Pinterest's implementation of a unified lightweight scoring system for its home feed recommendations using a two-tower architecture.

The article discusses the implementation of HNSW (Hierarchical Navigable Small World graphs) in Pinterest's Manas search engine for real-time embedding-based retrieval.

The article discusses the improvements made to Pinterest's recommendation system, Pixie, through the implementation of a lightweight ranking model.

The article discusses Pinterest's development of a visual skin tone model aimed at enhancing inclusivity in search and recommendation systems.

PinnerSage is a multi-modal user embedding framework developed by Pinterest to enhance recommendation systems by representing users with multiple embeddings that capture diverse interests.

The article discusses the development and implementation of Expanded Targeting for Pinterest ads, a feature designed to enhance ad performance by automatically connecting Promoted Pins to relevant ...

The article discusses PinText, a multitask text embedding system developed by Pinterest to improve the representation of textual data in their platform.

Pin2Interest (P2I) is a scalable system developed by Pinterest for classifying content by mapping over 200 billion Pins into a dynamic taxonomy of interests.

This article discusses the importance of annotations in Pinterest's visual discovery engine, focusing on how these keywords help in understanding text associated with Pins.

The article discusses the automation of the 'Shop the Look' feature on Pinterest, leveraging computer vision to enhance user engagement and streamline the shopping experience.