LLM-based Feature Extraction for Operational Optimization

Explore how large language models (LLMs) are revolutionizing feature extraction for operational optimization in this 27-minute conference talk by Xin Lian from Canva. Discover how LLMs like GPT-3.5 are being leveraged to understand user content more efficiently, reducing production time and operational costs. Learn from two real-world examples at Canva that demonstrate how LLM-based feature extraction has improved accuracy and velocity in content understanding for both consumers and creators. Gain insights into the practical applications of LLMs in natural language processing, computer vision, and MLOps from a senior machine learning engineer's perspective.