We're the engineers behind Pinterest, building a visual discovery engine powered by the latest in machine learning, 300 billion ideas and 500 million users.
Join us June 25 for our next Pinterest Labs Talk with UC Berkeley's Dawn Song for a discussion on AI & LLM agent advancements and key risks, moderated by Chuck Rosenberg. Register now 👉 pages.beamery.com/Pinterest/…
Real-time ad retrieval power-up 🚀
Hybrid inference combines sequential transformers context to blend high-intent signals with long-term user history. Result: 3x–10x Recall@K, ~300% higher median candidate relevance, and measurable ROAS lift. medium.com/pinterest-enginee…
New on the Blog: For large-scale ML systems, moving less data can be just as important as computing faster. This post shares how we reduce unnecessary feature-transfer overhead to improve network efficiency in production.
Read more: medium.com/pinterest-enginee…
From Determinantal Point Processes to sliding spectrum decomposition, here's how our team evolved multi-objective optimization at Pinterest to balance what people engage with and what keeps the experience inspiring. medium.com/pinterest-enginee…
At Pinterest, we solved the expensive, manual setup of Android performance metrics by integrating Visually Complete logic into the BaseSurface class.
Our "All-In-One Solution" story 🧵👇
This automatically measures User Perceived Latency on over 60 surfaces, saving two engineer-weeks per feature and encouraging platform-wide performance optimization. Learn more: medium.com/pinterest-enginee…
At Pinterest, we built a full MCP ecosystem — complete with a central registry, domain-specific servers & integrations across IDEs, chat & AI agents. We’re driving more than 66K monthly invocations & saving 7K engineering hours each month, all with security embedded from day one.
Text-to-SQL sounds simple until your warehouse has 100,000 tables.
At Pinterest, we built an analytics agent that goes beyond keyword matching by learning from analyst behavior, understanding intent, and incorporating signals like freshness and table quality.
New Blog 📌 Our engineers share how they unified two previously separate engagement models into a single architecture while preserving surface-specific nuance and improving efficiency at scale.
Learn more: medium.com/pinterest-enginee…
Introducing PinLanding, a production‑oriented pipeline for shopping collection generation at web scale. Multimodal AI, CLIP‑style models, LLM‑as‑judge, Ray, Spark, ANN search… all in one production system.
Deep dive: medium.com/pinterest-enginee…
Our new blog details how user surveys power a visual quality signal in our recommender systems, helping people find content that truly inspires them. Learn more: medium.com/pinterest-enginee…
How did the team reduce Android testing build times by more than 36%? Introducing PinTestLab, our in-house Android E2E testing solution. Learn more: medium.com/pinterest-enginee…
Our Pinterest Machine Learning Day Conference is one week away 🚨 Register now to join us on Nov 6 and discover what it’s like to build a platform for more than 570 million monthly active users. events.ringcentral.com/event…
Our Pinterest Machine Learning Day Conference is one week away 🚨 Register now to join us on Nov 6 and discover what it’s like to build a platform for more than 570 million monthly active users. events.ringcentral.com/event…
From inspiration to realization, learn how our team moved beyond understanding interests to comprehending long-term goals of Pinterest users by introducing user journeys as the foundation for recommendations. medium.com/pinterest-enginee…