Ray

1. Ray overview

   The Ray ecosystem of core, Data, Train, Tune, Serve, and RLlib libraries.

2. Ray Core key concepts

   Tasks, actors, objects, placement groups, and the distributed model.

3. Tasks

   Remote functions, futures, ray.get, and nested tasks.

4. Actors

   Stateful remote workers, named actors, and concurrency.

5. Objects and object store

   Object references, shared-memory store, serialization, and spilling.

6. Scheduling and resources

   Resource requirements, GPUs, placement groups, and scheduling strategies.

7. Fault tolerance

   Retries, actor restarts, object reconstruction, and node failures.

8. Ray Data

   Scalable data loading and batch transformation for ML pipelines.

9. Ray Train

   Distributed training with ScalingConfig, checkpointing, and fault handling.

10. Ray Tune

    Hyperparameter tuning with search algorithms and schedulers.

11. Ray Serve

    Model serving with deployments, autoscaling, and composition.

12. Clusters and deployment

    Running Ray clusters on Kubernetes and cloud providers, and autoscaling.

13. Observability

    Dashboard, metrics, logging, and debugging distributed applications.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.