Training Recommender System Models for Efficiency and Scale | Saurabh Vishwas Joshi (Pinterest) | Episode 2: Granica Tech Talks
Super excited to welcome Saurabh Vishwas Joshi, ML Engineering Lead at Pinterest, to share real-world insights on training Recommender System models for efficiency and scale!
Recommender systems are ubiquitous, powering personalized experiences in various industries. From suggesting movies on streaming platforms and recommending products in e-commerce to personalizing content on social media and optimizing search results, recommender systems drive user engagement and revenue.
In this tech talk we will dive into optimizing ML efficiency at scale within recommender systems, focusing on techniques to improve training performance and scalability using Ray, a distributed computing framework. Attendees will learn about critical approaches for enhancing recsys efficiency and throughput when handling large datasets.
We’ll cut through the hype and dive straight into what really matters:
Scaling Challenges in Recommender System Training
Core Obstacles in Model Efficiency
Ray’s Role in Distributed Scaling
Effective Infrastructure Setup
Pipeline and Data Format Optimizations
This is for ML and AI engineers and researchers who want the unfiltered truth about building large-scale ML systems. No fluff—just facts and actionable insights. You'll walk away with real knowledge and join a great conversation.
Limited seats.
Granica HQ, Mountain View
Pizza and beers on us.
6p-630p: Doors open & Networking
630p-730p: Talk
730p-8p: Networking
RSVP now to save your spot!
—
This is Episode 2 of Granica Tech Talks, a new series featuring distinguished speakers discussing the most important topics in AI, ML, and distributed systems. Join us for insightful conversations and actionable takeaways.