skills.rest
infantesromeroadrian
infantesromeroadrianCOMMUNITY
Explore all skills by this author →
Website𝕏@infantes_adrianGitHub

llm-engineering

Configure PEFT fine-tuning, guardrails, and vLLM serving workflows.

Updated Aug 27, 2026
Search Tags:#llm#alignment#guardrails#lora#peft#qlora#serving
One-click install
npx skills add https://github.com/infantesromeroadrian/arca-claude-code --skill llm-engineering-infantesromeroadrian
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: llm-engineering
Source: https://github.com/infantesromeroadrian/arca-claude-code/tree/main/skills/llm-engineering
Command: npx skills add https://github.com/infantesromeroadrian/arca-claude-code --skill llm-engineering-infantesromeroadrian

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This guide provides a comprehensive overview of LLM engineering, covering fine-tuning (PEFT), structured outputs, guardrails, alignment, and production serving to streamline building robust, production-ready LLM systems.

Core Features & Use Cases

  • Fine-Tuning with PEFT (LoRA, QLoRA) to tailor models efficiently for domain tasks.
  • Structured outputs and alignment tools (Instructor, DPO/RLHF) to improve reliability and safety.
  • Guardrails integration (NeMo Guardrails, Guardrails AI) to enforce content and behavior policies.
  • Production serving with vLLM for scalable, OpenAI-compatible APIs.
  • End-to-end workflows combining training, evaluation, and deployment tasks.

Quick Start

Follow the guide to configure an end-to-end LLM engineering workflow in your project.

Frequently Asked Questions about llm-engineering

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I configure PEFT with LoRA or QLoRA for fine-tuning large language models?▼

PEFT fine-tuning with LoRA or QLoRA tailors models efficiently for domain tasks. You configure parameter-efficient methods using TRL SFTTrainer to adapt large language models without updating all weights, reducing computational overhead.

What is the best way to enforce content policies using NeMo Guardrails?▼

NeMo Guardrails enforces content and behavior policies for LLMs. Integrating guardrails within your inference pipeline restricts unsafe outputs, ensuring your production models adhere to specified safety and topical constraints.

Can I use vLLM to serve an OpenAI-compatible API for my fine-tuned models?▼

vLLM provides production serving with scalable, OpenAI-compatible APIs. You can deploy your fine-tuned models using vLLM to handle high-throughput inference requests, exposing them through standard API endpoints.

How does DPO or RLHF alignment improve structured outputs?▼

DPO and RLHF alignment improve reliability and safety by optimizing model preferences. Combined with structured output tools like Instructor, these alignment techniques ensure models generate predictable, schema-compliant responses.

When should I use QLoRA instead of standard LoRA for fine-tuning?▼

Use QLoRA instead of standard LoRA when GPU memory is highly constrained. QLoRA quantizes base model weights to 4-bit before applying LoRA adapters, significantly lowering memory requirements for fine-tuning large models.

Related Skills

Explore More Software Engineering Skills→
SriRamkunamsettySriRamkunamsetty

fine-tuning-with-trl

Fine-tune language models with TRL RLHF workflows for human preference alignment.

Community
Advanced
ovachieverovachiever
52

serving-llms-vllm

Deploy OpenAI-compatible LLM APIs with vLLM for high-throughput inference.

Community
Advanced
Clay-HHKClay-HHK
1

outlines

Generate structured JSON outputs from local language models using Pydantic schemas and grammar-constrained sampling.

Community
Intermediate
Supporter09Supporter09

peft-fine-tuning

Train lightweight adapters for memory-efficient LLM fine-tuning.

Community
Advanced