24 articles
This article presents testing methods to measure and improve AI agent skill invocation reliability using Pinterest's internal agents and Claude Code.
This paper presents a Contextual Sequential Two-Tower Model for Pinterest ads that integrates real-time context into sequential recommender systems.
Pinterest optimized ML serving network efficiency by implementing Feature Trimmer to reduce bandwidth bottleneck.
Pinterest built an ML model optimizing shopping conversions by addressing sparse offsite conversion events.
Pinterest's MIQPS algorithm automatically learns which URL parameters affect content identity, enabling efficient deduplication across millions of merchant URLs at scale.
Pinterest engineers debugged why Ray-based ML training jobs were crashing with intermittent network connectivity issues on Kubernetes clusters backed by AWS EC2.
Pinterest shares their technique of request-level deduplication to manage infrastructure costs when scaling recommendation systems with 100x increased model parameters.
Pinterest's approach to automatically measuring user-perceived latency (Visually Complete) on Android surfaces by embedding measurement logic into base UI classes.
This article discusses Pinterest's evolution of the multi-objective optimization layer in their home feed recommendation system.
Pinterest describes how they built a production MCP (Model Context Protocol) ecosystem to enable AI agents to safely automate engineering tasks.
Pinterest evolved its Text-to-SQL system into a production Analytics Agent using unified context-intent embeddings and governance-aware ranking to serve 100,000+ tables and 2,500+ analysts.
Pinterest unified three separate ads engagement models for Home Feed, Search, and Related Pins into one shared architecture.
Pinterest investigates the online–offline discrepancy in L1 CVR models in their ads funnel.
Pinterest's Piqama is a generic quota management ecosystem that handles the full lifecycle of resource quotas across Big Data Processing and Online Services.
This post describes Pinterest's Auto Memory Retries feature for Apache Spark, which automatically retries OOM-failed tasks on larger executors to reduce failures and resource waste.
Pinterest introduced a GPU-served two-tower model using MMOE-DCN architecture for lightweight ads engagement prediction.
Pinterest describes its next-generation database ingestion framework built on CDC, Kafka, Flink, Spark, and Iceberg to replace legacy batch-based pipelines.
Spotify's ads team describes how they re-architected their serving stack to replace the Two-Tower model with more expressive neural networks capable of deep feature interactions.
Pinterest's Ads team developed transformer-based behavioral sequence models to improve ad candidate generation using users' offsite activity history.
Pinterest introduces PinLanding, a production pipeline that uses multimodal AI to automatically generate shopping collections from billions of catalog items.
