This course provides a technical guide to designing, building, and managing the hardware and software stack for machine learning applications. You will learn to evaluate compute requirements, compare on-premise and cloud solutions, and implement strategies for performance and cost optimization. The material covers hardware selection including CPUs and GPUs, containerization with Docker and Kubernetes for ML, and techniques for efficient model training and deployment. This is a practical course for engineers responsible for the operational aspects of AI systems.
Prerequisites ML concepts and systems knowledge
Level:
Hardware Selection
Evaluate and select appropriate compute hardware, including CPUs, GPUs, and specialized accelerators for different AI workloads.
Infrastructure Design
Design infrastructure solutions for both on-premise and cloud environments based on performance and budget requirements.
Containerization and Orchestration
Use Docker and Kubernetes to create reproducible ML environments and orchestrate distributed training and inference workloads.
Performance Optimization
Apply techniques such as distributed training, mixed-precision, and model quantization to improve the efficiency of AI systems.
Cost Management
Analyze and manage the costs associated with AI infrastructure, implementing strategies to optimize spending in both cloud and on-premise setups.