In-person
Virtual
In-person
From GPUs to MLOps: building a platform for multimodal AI applications
What is the role of the platform engineer in an AI company building foundation models? Explore cutting-edge AI/ML infrastructure with a platform engineer from a startup building multimodal oncology foundation models. This talk takes us from the fundamental GPU requirements and infrastructure as code to building full ML platforms on Kubernetes and cloud-native MLOps tools.
- Insights into the unique challenges and solutions for ML infrastructure in training multimodal foundation models. GPU requirements, integrating big data technologies, HPC, and CUDA for distributed training.
- Tools and principles for a cloud-native ML platform, including infrastructure-as-code, containerization, and orchestration
- How to identify and address gaps in ML workflows. Learn to balance the needs of data scientists, ML engineers, and data engineers while maintaining a robust, scalable, and secure ML infrastructure.
