planned-toc

Chapter 1: What is Ray & where can it be used & how it compares in the ecosystem
○	Distributed/Magically Scaleable Python
○	Celery, Spark, Airflow, Dask, multiprocessing, kubeflow
●	Chapter 2: Installing Ray & a hello world
○	How to install ray locally
○	References to the cloud provider options & K8s in the appendix
○	Computing 420! with ray
●	Chapter 3: Understanding the basics of ray
○	The different paradigms of programming in ray & how to choose
○	Execution model
○	Fault tolerance overview & implications
○	Handling Dependencies
●	Chapter 4: Stateless* Serverless
○	Remote functions & tasks
○	General APIs
○	Handling user requests
●	Chapter 5: Batch Data-Parallel processing
○	Loading data
○	Data parallel operations
○	Data sources in detail
●	Chapter 6: Actors (going beyond stateless)
○	Basics
○	Actors for ML
○	Scaling your actors
○	Where to store your state & implications for fault tolerance
●	Chapter 7: Streaming
○	Sources & sinks
○	Stateful streaming with Ray
●	Chapter 8: Higher-level building blocks
○	Pools, etc. - letting you plug in random shit
○	DataFrames (almost) w/Arrow
○	DataFrames w/Koalas w/Spark (ugh)
○	ML: tune & reinforcement learning
●	Chapter 9: Integrated libraries - Using, limitations, etc.
○	Pytorch
○	Pandas
●	Chapter 10: Porting existing code
○	Understanding implications with state
○	Finding components suitable for distributed computation
●	Chapter 11: Using Ray on GPUs
○	Rapids etc.
●	Appendix A: Notebooks
●	Appendix B: Deploying Ray
○	Running Ray on Kubernetes
○	Ray on Clouds
●	Appendix C: Debugging