Ray
Ray distributed computing framework. Use for scaling ML.From its SKILL.md
npx -y skills add G1Joshi/Agent-Skills --skill rayAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 12 stars12 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
1.1 KB, 248 tokens by cl100k_base, as published. Nobody here has run it
Ray
Ray is the compute layer for AI. It powers ChatGPT training and massive scale workloads. v3.0 (2025) improves efficiency and adds an MCP Server for agents.
When to Use
- Distributed Training: Scaling PyTorch across 100 GPUs.
- Ray Serve: Serving LLMs with high throughput (vLLM integration).
- Hyperparameter Tuning: Ray Tune is the industry standard.
Core Concepts
Actors & Tasks
- Task: Stateless function (like Lambda).
- Actor: Stateful class (like a microservice).
Object Store
Shared memory across the cluster means zero-copy data sharing.
Best Practices (2025)
Do:
- Use
ray.data: For streaming massive datasets into trainers. - Use KubeRay: The Kubernetes operator for managing Ray clusters.
- Use Ray Serve: It supports "Model Composition" (chaining models).
Don't:
- Don't use for simple scripts: The overhead of starting a Ray cluster is 5-10s.
References
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.