AI LLM Development & Engineering

About

This skill provides a comprehensive framework for building and scaling robust LLM applications using modern industry standards. It offers deep technical guidance on choosing between RAG, fine-tuning, and agentic workflows, supported by decision matrices, production-ready templates, and rigorous evaluation patterns. From optimizing high-throughput serving with vLLM to implementing multi-layered safety guardrails, it equips developers with the necessary patterns to move from prototype to production-grade AI systems while avoiding common anti-patterns like hallucination and context overload.

Key Features

• Production checklists for pre-deployment validation and safety
• Copy-paste templates for CoT, ReAct, and multi-agent workflows
• 22 GitHub stars
• Advanced evaluation patterns including LLM-as-judge and multi-metric testing
• Architecture decision matrices for RAG, agents, and fine-tuning
• Inference optimization guidance for vLLM, quantization, and high-throughput serving

Use Cases

• Setting up LLMOps pipelines for model monitoring, drift detection, and evaluation
• Designing and implementing a hybrid RAG pipeline with reranking and contextual retrieval
• Developing agentic workflows using LangGraph, CrewAI, or reflection patterns