How BMW Scales Automotive AI Workloads with the Ray Framework | Ray Summit 2025

At Ray Summit 2025, Thomas Riedl from BMW shares how the company is building scalable, safety-critical AI systems through the Connected AI Platform—a software-engineering–driven MLOps stack that accelerates the development and deployment of machine learning models across both cloud and in-vehicle environments. He begins by outlining the rising demands of modern automotive AI, where reliability, reproducibility, and scale are essential for real-world deployment. A central pillar of BMW’s approach is its adoption of Ray, the distributed computing framework from Anyscale, which enables efficient scaling for both classical ML workloads and large GenAI models. Thomas demonstrates how Ray powers: Distributed fine-tuning and inference of Large Language Models with strong reproducibility guarantees Robust, cloud-scale training workflows for safety-critical automotive use cases Flexible, dynamic resource management across heterogeneous compute environments The foundation for future multimodal workflows—spanning video, text, and sensor data Major improvements in big-data processing, enabling faster iteration and more efficient model development He also shares the engineering practices that make these systems production-ready, including platform architecture decisions, operational guardrails, and lessons learned from integrating distributed AI into enterprise pipelines. Attendees will walk away with practical insights into building scalable automotive AI systems with Ray—and how BMW’s Connected AI Platform empowers rapid innovation while meeting the stringent reliability requirements of the automotive domain. Liked this video? Check out other Ray Summit breakout session recordings    • Ray Summit 2025 - Breakout Sessions   Subscribe to our YouTube channel to stay up-to-date on the future of AI!    / anyscale   🔗 Connect with us: LinkedIn:   / joinanyscale   X: https://x.com/anyscalecompute Website: https://www.anyscale.com/