What types of GPUs can I access with this skill?

This skill supports a wide range of NVIDIA GPUs including the H100, A100 (40GB and 80GB variants), L40S, L4, A10G, and T4.

Do I need Docker or Kubernetes to use this skill?

No, Modal handles containerization and orchestration automatically directly from your Python code, eliminating the need for Dockerfiles, YAML, or cluster management.

How does the autoscaling work?

Modal functions can scale from zero to hundreds of instances instantly based on demand. You can configure parameters like min_containers to eliminate cold starts or max_containers to control costs.

Can I store persistent data between function runs?

Yes, this skill utilizes Modal Volumes for high-performance persistent file storage and Modal Dicts or Queues for distributed key-value storage and task management.

Modal Cloud Deployment

Name: Modal Cloud Deployment
Author: ferdousbhai

byferdousbhai

•

云基础设施

Deploys serverless Python applications with instant GPU access, autoscaling, and zero-configuration infrastructure.

This skill enables Claude to architect and deploy high-performance Python applications to the Modal platform without writing YAML or managing Kubernetes. It provides specialized knowledge for setting up serverless containers, configuring various GPU types (from T4 to H100), managing persistent volumes, and deploying scalable web endpoints. Whether you are running batch data processing, fine-tuning LLMs, or hosting high-traffic APIs, this skill ensures best practices for image building, secret management, and observability using modern tools like Logfire and uv.

主要功能

01High-performance image builds with uv_sync for faster cold starts

022 GitHub stars

03Zero-config serverless containerization and deployment

04On-demand GPU acceleration (H100, A100, L4) for ML workloads

05Persistent storage via Modal Volumes and distributed state management

06Automated scaling from zero to hundreds of concurrent containers

使用场景

01Running large-scale parallel data processing and batch jobs

02Deploying LLM inference APIs with A100/H100 GPU acceleration

03Scheduling recurring cron jobs for automated reports or system maintenance

主要功能

01High-performance image builds with uv_sync for faster cold starts

022 GitHub stars

03Zero-config serverless containerization and deployment

04On-demand GPU acceleration (H100, A100, L4) for ML workloads

05Persistent storage via Modal Volumes and distributed state management

06Automated scaling from zero to hundreds of concurrent containers

使用场景

01Running large-scale parallel data processing and batch jobs

02Deploying LLM inference APIs with A100/H100 GPU acceleration

03Scheduling recurring cron jobs for automated reports or system maintenance