Helios GPU Cloud Skills

Agent skills for deploying and training AI models on Helios GPU Kubernetes clusters. Drop your kubeconfig, start finetuning.

GPU Infrastructure

Skills that get you from zero to training on NVIDIA B200 GPUs.

• unsloth-studio-deploy — Deploy Unsloth Studio on a Helios K8s cluster with B200 GPUs. Handles GPU enablement, namespace setup, persistent storage, and Studio launch. Just provide your kubeconfig.

  npx skills@latest add HeliosEnergy/skills/unsloth-studio-deploy
  

Coming Soon

• vllm-deploy — Deploy vLLM inference server on Helios B200 GPUs with auto-scaling.

• multi-node-training — Set up multi-node distributed training with DeepSpeed or FSDP across Helios GPU nodes.

• model-serving-pipeline — End-to-end pipeline: finetune with Unsloth, quantize to GGUF, deploy with Ollama or vLLM on Helios.

• gpu-monitoring — Deploy Prometheus + Grafana GPU monitoring stack (DCGM exporter) on Helios clusters.

---

Built for Civo Helios GPU Kubernetes clusters with NVIDIA B200 GPUs.