hf-cloud-sagemaker-deployment-planner

What problem does it solve? Deploying a model to Amazon SageMaker requires choosing between real-time, serverless, async, batch, and Bedrock pathways, each with different cost, latency, and quota constraints. This Skill guides that decision with minimal questions and hands off to specialized implementation skills. ## Core Features & Use Cases - Pathway Selection: Compares real-time endpoints, serverless inference, async inference, batch transform, and Bedrock Custom Model Import against traffic shape, latency tolerance, and model type. - Quota-Aware Instance Recommendation: Checks SageMaker service quotas per region before recommending GPU instance families like ml.g5, ml.g6, or ml.g6e to avoid ResourceLimitExceeded failures. - Workflow Orchestration: Coordinates handoffs to context discovery, IAM preflight, image selection, and production deployment skills. - Use Case: A user wants to deploy a fine-tuned LLM. The Skill asks only what model and how often it will be called, recommends a real-time endpoint on ml.g5.xlarge after verifying quota, then defers implementation to downstream skills. ## Quick Start Ask the assistant to plan a SageMaker deployment for your HuggingFace model, describing roughly how often it will be called.