ray-project/ray — reverse-engineered prompt

Build me a Python framework that helps people run AI and regular Python workloads on more than one machine without rewriting everything. I want someone to be able to start on their laptop, then move the same code to a bigger cluster or cloud setup when the job gets too large.

It should let users send normal Python functions to run in parallel, keep long running worker objects alive, and share results between workers. Include simple tools for common AI work like processing big datasets, distributed model training, tuning model settings, reinforcement learning, and serving trained models as an API.

Please include a basic dashboard or monitoring view so users can see what jobs are running, what failed, and how the cluster is doing. Make the install and first example really simple, like a quick pip install and a tiny script that proves parallel work is happening. Look up current docs online if you need to.