Free resources
Guides and templates for reliable AI systems
Use these practical checklists and templates without signing up. Each one comes from production work, including the parts that needed a rewrite.
- CheatsheetEval-First AI Engineering
Eval metric selection checklist
The five-step procedure for testing whether a metric actually separates a good build from a regressed one, before it goes in your CI gate.
Coming soon MCP server starter (TypeScript)
A minimal, production-shaped MCP server template: per-tool timeouts, a circuit breaker, and schema versioning wired in from the start.
Coming soon- CheatsheetProduction RAG & MLOps
RAG index health queries
The four SQL queries that diagnose most retrieval regressions — freshness, embedding-model drift, source-mix shift, and empty-filter rate.
Coming soon - TemplateEval-First AI Engineering
LLM-as-judge rubric template
A per-claim decomposition rubric that holds up better than a 1–5 Likert score — the shape that survived the 45-metric teardown.
Coming soon
Newsletter
Get new resources first
New guides and templates land in the newsletter before they land here.