Enabling a Scalable ML Driven Search Experience at Coursera

Join us as we share Coursera’s experience of migrating to OpenSearch, impacting over 3 million monthly active users. In this presentation, we’ll cover the key decisions and challenges we encountered, including the integration of vector search capabilities, and our choice to leverage Coursera’s in-house recommendation system (RecSys). We’ll also discuss how we conducted offline and online evaluation experiments to ensure a smooth transition and maintain search quality.

We’ll provide a detailed look at our approach to light and heavy ranking, as well as the load testing and performance tuning that were essential to optimizing our search infrastructure. Learn about the results of our migration efforts, including improvements in search accuracy, system performance, and user satisfaction. This session aims to offer practical insights and lessons learned for anyone interested in search technology and large-scale system migrations.

Details

Tuesday, September 24 1:00pm-1:30pm in Continental BR 1-3

Track: Operating OpenSearch

Speakers

J.R. Maingat photograph

J.R. Maingat

Senior Software Engineer at Coursera