modal-serverless-gpu

Quick Review

Excellent serverless GPU skill with comprehensive coverage of Modal platform capabilities. The description clearly conveys when to use Modal vs alternatives. Task knowledge is outstanding with complete code examples for common patterns (inference endpoints, batch jobs, web APIs, parallel processing). Structure is clean with good use of tables, clear sections, and appropriate delegation to reference files. Novelty is strong as Modal's Python-native infrastructure-as-code and GPU orchestration would require significant token investment for a CLI agent to discover and implement correctly. Minor improvement areas: could add more specific cost optimization patterns in main file and perhaps a decision tree for GPU selection. Overall, this is a production-ready skill that significantly reduces the complexity of deploying ML workloads on serverless infrastructure.