AI infrastructure courses and concepts
AI infrastructure is the hardware, systems and working practices that AI depends on: specialized chips, the services that deliver a trained model's output to the people and software using it, the stores that hold the information a model draws on, and the practices that keep a model working after it goes live. Training a large model is expensive, but inference, the model producing an output, costs money every time it runs, so the cost of a model keeps growing with how much it is used. Learning it means following a model past training, through deployment, monitoring, and retraining when its accuracy drops.
AI infrastructure courses
Advanced Machine Learning Operations
Databricks
Building Language Models on AWS
AWS Skill Builder
Google Cloud AI Infrastructure
Google Skills
Machine Learning at Scale
Databricks
Machine Learning Model Deployment
Databricks
Operationalize generative AI applications (GenAIOps)
Microsoft Learn
Operationalize machine learning models (MLOps)
Microsoft Learn
Professional Machine Learning Engineer Certification
Google Skills
The Art of Compressing LLMs: Pruning, Distillation, and Quantization
NVIDIA Deep Learning Institute
AI infrastructure terms explained
Most of these terms fall into four groups: the chips AI runs on (graphics, tensor and neural processing units), where and how a trained model runs (inference, model deployment, edge AI, quantization), keeping a model working once it is live (machine learning operations, model monitoring, model drift), and stores of information that AI systems search or draw on (vector databases, knowledge graphs, knowledge bases).
Develop AI is where AI infrastructure courses sit, alongside courses on building and training models and on building applications with them.