tools-only vastai-reliability-patterns
Implement reliability patterns including checkpointing, graceful degradation, and fault tolerance for spot GPU workloads
Directory Structure
vastai-reliability-patterns/
├── SKILL.md
└── examples/
└── fault_tolerant_trainer.py
File Descriptions
| File | Type | Purpose |
|---|---|---|
| SKILL.md | Markdown | Reliability engineering patterns for GPU cloud with spot instance considerations |
| examples/fault_tolerant_trainer.py | Python | Training script with checkpointing, spot interruption handling, and automatic resume |
Summary
Category: Advanced Target Audience: Platform architects, SREs building resilient systems, senior ML engineers Trigger Phrases: "vast checkpointing", "vast fault tolerance", "vast spot handling", "vast reliability", "vast training resume", "vast graceful shutdown"
Version: 1.0.0 License: MIT Author: Jeremy Longshore jeremy@intentsolutions.io