high-performance-inference
High-Performance Inference
Optimize LLM inference for production with vLLM 0.14.x, quantization, and speculative decoding.
vLLM 0.14.0 (Jan 2026): PyTorch 2.9.0, CUDA 12.9, AttentionConfig API, Python 3.12+ recommended.
Overview
- Deploying LLMs with low latency requirements
- Reducing GPU memory for larger models
- Maximizing throughput for batch inference
- Edge/mobile deployment with constrained resources
- Cost optimization through efficient hardware utilization
Quick Reference
# Basic vLLM server
vllm serve meta-llama/Meta-Llama-3.1-70B-Instruct \
More from yonatangross/skillforge-claude-plugin
zustand-patterns
Reference for Zustand 5.x state management including slices, middleware, Immer, useShallow, persistence, selectors, and devtools integration. Documents 7 core patterns with TypeScript examples and anti-patterns. Use when building React state management with Zustand instead of Redux.
45domain-driven-design
Domain-Driven Design tactical patterns for complex business domains. Use when modeling entities, value objects, domain services, repositories, or establishing bounded contexts.
38doctor
OrchestKit doctor for health diagnostics. Use when running checks on plugin health, diagnosing problems, or troubleshooting issues.
37react-server-components-framework
Use when building Next.js 16+ apps with React Server Components. Covers App Router, Cache Components (replacing experimental_ppr), streaming SSR, Server Actions, and React 19 patterns for server-first architecture.
36skill-evolution
Tracks skill usage patterns, edit frequency, and success rates to suggest improvements and optimizations. Manages skill versioning with safe rollback capability and confidence scoring for suggestions. Use when reviewing skill performance, applying auto-suggested changes, or rolling back problematic versions.
36architecture-decision-record
Use this skill when documenting significant architectural decisions. Provides ADR templates following the Nygard format with sections for context, decision, consequences, and alternatives. Use when writing ADRs, recording decisions, or evaluating options.
36