The AI Engineer Stack

Claude Code skills for AI engineers cover the full production stack — from system architecture and model plumbing through to the observability that keeps things working once deployed. This stack was built for practitioners who ship real AI systems, not just notebook experiments.

What these skills do

RAG Architect

Design retrieval-augmented generation systems with the right chunking strategy, embedding model, vector store, and retrieval logic for your specific data and query patterns. Stops you from building the wrong architecture for your use case.

Agent Designer & Agent Workflow Designer

Structure multi-step agent systems with clear tool definitions, state management, and fallback handling. The workflow designer handles orchestration between agents — useful when you’re building systems where multiple agents hand off tasks to each other.

MCP Server Builder

Build Model Context Protocol servers that give Claude structured access to external tools and data sources. Covers authentication, tool schema design, and the common patterns for connecting AI agents to APIs, databases, and file systems.

Senior Prompt Engineer

Craft prompts that produce consistent, predictable outputs across edge cases. Covers chain-of-thought structuring, few-shot design, system prompt architecture, and testing prompt reliability at scale.

Senior ML Engineer & Senior Data Scientist

Get senior-level thinking on model selection, training decisions, experiment design, and the tradeoffs between different approaches. Useful when you’re making decisions that will be expensive to reverse.

Senior Data Engineer

Design data pipelines for AI workloads — feature stores, training data pipelines, streaming ingestion, and the schema decisions that make downstream ML work easier.

Senior Computer Vision

Architecture and implementation guidance for CV systems: model selection, preprocessing pipelines, inference optimization, and the annotation strategies that affect model quality downstream.

Observability Designer

Instrument AI systems with the right metrics, traces, and logs. Covers LLM-specific observability — latency, token usage, retrieval quality, hallucination detection — not just generic application monitoring.

Performance Profiler

Find and fix bottlenecks in AI inference pipelines, RAG retrieval, and agent execution. Covers profiling methodology, common hotspots in LLM-based systems, and the tradeoffs between latency and throughput.

Who this is for

• AI engineers building LLM-powered applications for production
• ML engineers moving models from research to deployed systems
• Full-stack developers adding AI features to existing products
• Platform engineers building internal AI tooling or agent infrastructure

Pair this stack with the developers audience page for more technical skills, or see the DevOps & Platform Stack if your bottleneck is infrastructure and deployment rather than AI architecture.