The automotive industry is a major global economic driver with significant market capitalization and employment, but it faces stagnating car sales despite growing mobility demands.
Machine learning finds patterns in vast amounts of car data to create a virtuous cycle of better products and more users, similar to how the internet industry evolved.
Machine learning improves automated driving by reducing costs and time while increasing quality, and it also enhances the in-car experience through personalization.
A key challenge is managing the terabytes of data generated by modern cars daily and integrating expensive sensors without destroying thin profit margins.
The automotive industry must shift from complex legacy architectures and waterfall development to agile, data-driven processes that support the entire ML lifecycle.
ML models can learn incorrect correlations from data, as shown by husky/wolf and criminal-detection examples, highlighting the need for better interpretability for safe deployment.
Jobs that call for the skills explored in this talk.