Optimizing EV Charging Experience - Machine Learning for Accurate Charge Time Estimation

Learn how to develop and deploy machine learning models for accurate electric vehicle charge time estimation in this 38-minute conference talk from Databricks. Discover how Rivian Automotive overcame the limitations of traditional rule-based charging prediction methods by implementing an adaptive ML solution that improves accuracy by 10% while handling dynamic factors like environmental conditions, vehicle state, and charging infrastructure variables. Explore the comprehensive technical architecture including Unity Catalog for data governance, Delta Tables for scalable storage, Liquid Clustering for optimized data layout, and automated job schedulers for efficient data processing. Master the use of AutoML for accelerated model selection, MLflow for comprehensive experiment tracking and model versioning, and dedicated serving endpoints that enable A/B testing and real-time performance monitoring. Understand how this scalable ML framework was successfully integrated into vehicle control systems within months and now supports over 50,000 weekly charging sessions. Gain insights into live accuracy tracking methodologies, software-driven model blending techniques, and strategies for managing growing data ecosystems while maintaining system performance and enhancing both energy management efficiency and user experience in electric vehicle applications.