The Art of Hypothesis-Driven Development of ML-Powered Search - Haystack EU 2023
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only! Grab it Explore the intricacies of developing and optimizing ML-powered search systems for large-scale e-commerce platforms in this conference talk from Haystack EU 2023. Dive into the experiences and insights shared by Andrey Kulagin, Head of Machine Learning at Uzum Market, as he discusses the challenges and solutions in creating an effective search system for a rapidly growing marketplace with over 500,000 items. Learn about the evolution of their search pipeline, including sparse retrieval, spelling correction, typing suggestions, and machine learning models for ranking. Discover the mistakes made, principles learned, and strategies developed for streamlining hypothesis testing. Gain valuable knowledge on selecting appropriate metrics for search relevance, utilizing clickstream data for learning-to-rank models, and implementing effective AB testing methodologies. This talk provides essential insights for professionals working on large-scale search systems in e-commerce and beyond.