AI Engineering

Production-grade guidance for building AI agents, LLM systems, and data infrastructure. This skill covers the full AI engineering stack: from system prompt architecture to eval gates, from memory systems to fine-tuning pipelines, from RAG to multi-agent coordination.

Start Here

1. Identify the domain of your task using the Workflow Router below — load only relevant reference files.
2. For agent or prompt work: start with the system prompt architecture and tool design references before writing any prompt copy.
3. For quality and reliability: the eval system reference is required reading before shipping any agent capability change.
4. For data pipeline work: choose the processing model first (batch vs streaming, warehouse vs lakehouse).
5. Use templates/ for copy-paste starting points — system prompts and eval contracts have templates.
6. Scripts in scripts/ handle common operations; do not write boilerplate from scratch.

Core Rules