[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-9714":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":25,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":46,"readmeContent":47,"aiSummary":48,"trendingCount":16,"starSnapshotCount":16,"syncStatus":17,"lastSyncTime":49,"discoverSource":50},9714,"skypilot","skypilot-org\u002Fskypilot","skypilot-org","Run, manage, and scale AI workloads on any AI infrastructure. Use one system to access & manage all AI compute (Kubernetes, Slurm, 20+ clouds, on-prem).","https:\u002F\u002Fdocs.skypilot.co\u002F",null,"Python",10078,1092,74,127,0,2,18,107,11,44.12,"Apache License 2.0",false,"master",true,[27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"cloud-computing","cloud-management","cost-optimization","deep-learning","distributed-training","gpu","hyperparameter-tuning","job-queue","job-scheduler","llm-serving","llm-training","machine-learning","ml-infrastructure","ml-platform","mlops","multicloud","slurm","spot-instances","tpu","2026-06-12 02:02:11","\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fskypilot-org\u002Fskypilot\u002Fmaster\u002Fdocs\u002Fsource\u002Fimages\u002Fskypilot-wide-dark-1k.png\">\n    \u003Cimg alt=\"SkyPilot\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fskypilot-org\u002Fskypilot\u002Fmaster\u002Fdocs\u002Fsource\u002Fimages\u002Fskypilot-wide-light-1k.png\" width=55%>\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdocs.skypilot.co\u002F\">\n    \u003Cimg alt=\"Documentation\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-gray?logo=readthedocs&logoColor=f5f5f5\">\n  \u003C\u002Fa>\n\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Freleases\">\n    \u003Cimg alt=\"GitHub Release\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Frelease\u002Fskypilot-org\u002Fskypilot.svg\">\n  \u003C\u002Fa>\n\n  \u003Ca href=\"http:\u002F\u002Fslack.skypilot.co\">\n    \u003Cimg alt=\"Join Slack\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSkyPilot-Join%20Slack-blue?logo=slack\">\n  \u003C\u002Fa>\n\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Freleases\">\n    \u003Cimg alt=\"Downloads\" src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fskypilot\">\n  \u003C\u002Fa>\n\n\u003C\u002Fp>\n\n\u003Ch3 align=\"center\">\n    Manage all your AI compute\n\u003C\u002Fh3>\n\n\u003Cdiv align=\"center\">\n\n#### [🌟 **SkyPilot Demo** 🌟: Click to see a 1-minute tour](https:\u002F\u002Fdemo.skypilot.co\u002Fdashboard\u002F)\n\n\u003C\u002Fdiv>\n\n\nSkyPilot is a system to run, manage, and scale AI workloads on any AI infrastructure.\n\nSkyPilot gives **AI teams** a simple interface to run jobs on any infra.\n**Infra teams** get a unified control plane to manage any AI compute — with advanced scheduling, scaling, and orchestration.\n\n\u003Cpicture>\n  \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\".\u002Fdocs\u002Fsource\u002Fimages\u002Fskypilot-abstractions-long-2-dark.png\">\n  \u003Cimg src=\".\u002Fdocs\u002Fsource\u002Fimages\u002Fskypilot-abstractions-long-2.png\" alt=\"SkyPilot Abstractions\">\n\u003C\u002Fpicture>\n\n-----\n\n:fire: *News* :fire:\n- [Apr 2026] Introducing **GPU Compass**: One dashboard to browse, compare pricing, and launch across every GPU cloud. Try it at [**gpus.skypilot.co**](https:\u002F\u002Fgpus.skypilot.co).\n- [Apr 2026] **Research-Driven Agents**: Agents read arxiv papers before coding, landed 5 llama.cpp kernel fusions and +15% faster flash attention in ~3 hours for ~$29: [**blog**](https:\u002F\u002Fblog.skypilot.co\u002Fresearch-driven-agents\u002F), [**HackerNews**](https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=47706141)\n- [Mar 2026] **Scaling Karpathy's Autoresearch**: Autoresearch runs 1 experiment at a time. We gave it 16 GPUs and let it run in parallel: [**blog**](https:\u002F\u002Fblog.skypilot.co\u002Fscaling-autoresearch\u002F), [**HackerNews**](https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=47442435)\n- [Mar 2026] **How H Company Unlocked Online RL and Unified their AI Platform**: [**case study**](https:\u002F\u002Fhcompany.ai\u002Funlocking-online-rl-skypilot)\n- [Mar 2026] **SkyPilot v0.12** released: Slurm Support, Job Groups for RL, Agent Skill, Recipes, Pool Autoscaling for Batch Inference, 7x Data Mounting, and More: [**Release notes**](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Freleases\u002Ftag\u002Fv0.12.0)\n- [Mar 2026] **SkyPilot Agent Skills**: GPU access and job management for AI agents: [**docs**](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Fskill.html)\n- [Jan 2026] **Shopify case study**: Shopify runs all AI training workloads on SkyPilot: [**case study**](https:\u002F\u002Fshopify.engineering\u002Fskypilot)\n- [Dec 2025] **SkyPilot v0.11** released: Multi-Cloud Pools, Fast Managed Jobs, Enterprise-Readiness at Large Scale, Programmability. [**Release notes**](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Freleases\u002Ftag\u002Fv0.11.0)\n- [Dec 2025] Train **an agent to use Google Search** as a tool with RL on your Kubernetes or clouds: [**blog**](https:\u002F\u002Fblog.skypilot.co\u002Fverl-tool-calling\u002F), [**example**](.\u002Fllm\u002Fverl\u002F)\n\n## Overview\n\nSkyPilot **is easy to use for AI users**:\n- Quickly spin up compute on your own infra\n- Environment and job as code — simple and portable\n- Easy job management: queue, run, and auto-recover many jobs\n\nSkyPilot **makes Kubernetes easy for AI & Infra teams**:\n\n- Slurm-like ease of use, cloud-native robustness\n- Local dev experience on K8s: SSH into pods, sync code, or connect IDE\n- Turbocharge your clusters: gang scheduling, multi-cluster, and scaling\n\nSkyPilot **unifies multiple clusters, clouds, and hardware**:\n- One interface to use reserved GPUs, Kubernetes clusters, Slurm clusters, or 20+ clouds\n- [Flexible provisioning](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fauto-failover.html) of GPUs, TPUs, CPUs, with smart failover\n- [Team deployment](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Freference\u002Fapi-server\u002Fapi-server.html) and resource sharing\n\nSkyPilot **maximizes GPU fleet utilization**:\n* Autostop: automatic cleanup of idle resources\n* Binpacking: workload binpacking on shared clusters\n* Intelligent scheduler: automatically schedule on the most available infra\n\nSkyPilot supports your existing GPU, TPU, and CPU workloads, with no code changes.\n\nInstall with uv ([also supported](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Finstallation.html): pip, nightly, from source)\n```bash\n# Choose your clouds:\nuv pip install \"skypilot[kubernetes,aws,gcp,azure,oci,nebius,lambda,runpod,fluidstack,paperspace,cudo,ibm,scp,seeweb,shadeform,verda]\"\n```\n\nTo use SkyPilot directly with your agent (Claude Code, Codex, etc.), install the [SkyPilot Skill](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Fskill.html). Tell your agent:\n```\nFetch and follow https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Fblob\u002FHEAD\u002Fagent\u002FINSTALL.md to install the skypilot skill\n```\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fsource\u002F_static\u002Fintro.gif\" alt=\"SkyPilot\">\n\u003C\u002Fp>\n\nCurrent supported infra: Kubernetes, Slurm, AWS, GCP, Azure, OCI, CoreWeave, Nebius, Lambda Cloud, RunPod, Fluidstack,\nCudo, Digital Ocean, Paperspace, Cloudflare, Samsung, IBM, Vast.ai, VMware vSphere, Seeweb, Prime Intellect, Shadeform, Verda Cloud, VastData, Crusoe.\n\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fskypilot-org\u002Fskypilot\u002Fmaster\u002Fdocs\u002Fsource\u002Fimages\u002Fcloud-logos-dark.png\">\n    \u003Cimg alt=\"SkyPilot\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fskypilot-org\u002Fskypilot\u002Fmaster\u002Fdocs\u002Fsource\u002Fimages\u002Fcloud-logos-light.png\" width=85%>\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\n\u003C!-- source xcf file: https:\u002F\u002Fdrive.google.com\u002Fdrive\u002Ffolders\u002F1S_acjRsAD3T14qMeEnf6FFrIwHu_Gs_f?usp=drive_link -->\n\n\n## Getting started\n\n[Install SkyPilot](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Finstallation.html) in 1 minute. Then, launch your first cluster in 2 minutes in [Quickstart](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Fquickstart.html).\n\nSkyPilot is BYOC: Everything is launched within your cloud accounts, VPCs, and clusters.\n\n## Benefits of SkyPilot on Kubernetes\n\nSkyPilot makes Kubernetes AI-native.\n\nIt turbocharges your existing Kubernetes clusters by **accelerating AI\u002FML velocity**:\n\n- AI-friendly interface to launch jobs and deployments\n- Much simplified interactive dev for K8s (SSH \u002F sync code \u002F connect IDE to pods)\n\n...and **optimizing GPU scheduling, utilization, and scaling**:\n\n- Advanced scheduling: Gang scheduling, multi-node jobs, and queueing\n- Multi-cluster support: Bring all your clusters under one control plane\n- Multi-cloud support: One consistent interface to manage many providers\n\nSee [SkyPilot vs Vanilla Kubernetes](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Freference\u002Fkubernetes\u002Fskypilot-and-vanilla-k8s.html) and this [blog post](https:\u002F\u002Fblog.skypilot.co\u002Fai-on-kubernetes\u002F) for more details.\n\n## SkyPilot in 1 minute\n\nA SkyPilot task specifies: resource requirements, data to be synced, setup commands, and the task commands.\n\nOnce written in this [**unified interface**](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Freference\u002Fyaml-spec.html) (YAML or Python API), the task can be launched on any available infra (Kubernetes, Slurm, cloud, etc.).  This avoids vendor lock-in, and allows easily moving jobs to a different provider.\n\nPaste the following into a file `my_task.yaml`:\n\n```yaml\nresources:\n  accelerators: A100:8  # 8x NVIDIA A100 GPU\n\nnum_nodes: 1  # Number of VMs to launch\n\n# Working directory (optional) containing the project codebase.\n# Its contents are synced to ~\u002Fsky_workdir\u002F on the cluster.\nworkdir: ~\u002Ftorch_examples\n\n# Commands to be run before executing the job.\n# Typical use: pip install -r requirements.txt, git clone, etc.\nsetup: |\n  cd mnist\n  pip install -r requirements.txt\n\n# Commands to run as a job.\n# Typical use: launch the main program.\nrun: |\n  cd mnist\n  python main.py --epochs 1\n```\n\nPrepare the workdir by cloning:\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fpytorch\u002Fexamples.git ~\u002Ftorch_examples\n```\n\nLaunch with `sky launch` (note: [access to GPU instances](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fcloud-setup\u002Fquota.html) is needed for this example):\n```bash\nsky launch my_task.yaml\n```\n\nSkyPilot then performs the heavy-lifting for you, including:\n1. Find the cheapest & available infra across your clusters or clouds\n2. Provision the GPUs (pods or VMs), with auto-failover if the infra returned capacity errors\n3. Sync your local `workdir` to the provisioned cluster\n4. Auto-install dependencies by running the task's `setup` commands\n5. Run the task's `run` commands, and stream logs\n\nSee [Quickstart](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Fquickstart.html) to get started with SkyPilot.\n\n## Runnable examples\n\nSee [**SkyPilot examples**](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Fdocs-examples\u002Fexamples\u002Findex.html) that cover: development, training, serving, LLM models, AI apps, and common frameworks.\n\nLatest featured examples:\n\n| Task | Examples |\n|----------|----------|\n| Training | [Verl](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fverl.html), [Finetune Llama 4](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fllama-4-finetuning.html), [TorchTitan](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Ftorchtitan.html), [PyTorch](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fgetting-started\u002Ftutorial.html), [DeepSpeed](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fdeepspeed.html), [NeMo](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fnemo.html), [Ray](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fray.html), [Unsloth](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Funsloth.html), [Jax\u002FTPU](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Ftpu.html), [OpenRLHF](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Ftraining\u002Fopenrlhf.html) |\n| Serving | [vLLM](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fserving\u002Fvllm.html), [SGLang](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fserving\u002Fsglang.html), [Ollama](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fserving\u002Follama.html) |\n| Models | [DeepSeek-R1](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fdeepseek-r1.html), [Llama 4](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fllama-4.html), [Llama 3](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fllama-3.html), [CodeLlama](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fcodellama.html), [Qwen](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fqwen.html), [Kimi-K2](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fkimi-k2.html), [Kimi-K2-Thinking](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fkimi-k2-thinking.html), [Mixtral](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fmodels\u002Fmixtral.html) |\n| AI apps | [RAG](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fapplications\u002Frag.html), [vector databases](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fapplications\u002Fvector_database.html) (ChromaDB, CLIP) |\n| Common frameworks | [Airflow](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fframeworks\u002Fairflow.html), [Jupyter](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fframeworks\u002Fjupyter.html), [marimo](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Fexamples\u002Fframeworks\u002Fmarimo.html)  |\n\nSource files can be found in [`llm\u002F`](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Ftree\u002Fmaster\u002Fllm) and [`examples\u002F`](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Ftree\u002Fmaster\u002Fexamples).\n\n## Learn more\nTo learn more, see [SkyPilot Overview](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002Foverview.html), [SkyPilot docs](https:\u002F\u002Fdocs.skypilot.co\u002Fen\u002Flatest\u002F), and [SkyPilot blog](https:\u002F\u002Fblog.skypilot.co\u002F).\n\nSkyPilot adopters: [Testimonials and Case Studies](https:\u002F\u002Fblog.skypilot.co\u002Fcase-studies\u002F)\n\nPartners and integrations: [Community Spotlights](https:\u002F\u002Fblog.skypilot.co\u002Fcommunity\u002F)\n\nFollow updates:\n- [Slack](http:\u002F\u002Fslack.skypilot.co)\n- [X](https:\u002F\u002Ftwitter.com\u002Fskypilot_org)\n- [LinkedIn](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fskypilot-oss\u002F)\n- [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@skypilot-org)\n- [SkyPilot Blog](https:\u002F\u002Fblog.skypilot.co\u002F)\n\n## Questions and feedback\nWe are excited to hear your feedback:\n* For issues and feature requests, please [open a GitHub issue](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Fissues\u002Fnew).\n* For questions, please use [GitHub Discussions](https:\u002F\u002Fgithub.com\u002Fskypilot-org\u002Fskypilot\u002Fdiscussions).\n\nFor general discussions, join us on the [SkyPilot Slack](http:\u002F\u002Fslack.skypilot.co).\n\n## Contributing\nWe welcome all contributions to the project! See [CONTRIBUTING](CONTRIBUTING.md) for how to get involved.\n","SkyPilot 是一个用于在任何AI基础设施上运行、管理和扩展AI工作负载的系统。其核心功能包括通过统一控制面板管理Kubernetes、Slurm、20多个云平台及本地计算资源，支持分布式训练、超参数调优、GPU和TPU优化等高级调度与编排特性。该项目特别适合需要跨多种计算环境高效部署机器学习模型的企业或研究团队使用，能够显著提高资源利用率并降低成本。基于Python开发，采用Apache License 2.0开源许可。","2026-06-11 03:24:21","top_topic"]