AI Engineering Resources (mmfm.bearblog.dev)

This piece is a curated, opinionated roadmap of AI engineering resources designed to shorten the “trial-and-error” slog for practitioners moving from theory to production. It clarifies that the list targets AI engineering — practical, relatively mature practices around state‑of‑the‑art models — not frontier research, and emphasizes that most items teach “how to do X” but require hands‑on practice to stick. Key top-level recommendations include Chip Huyen’s AI Engineering book (good for broad skims; up‑to‑date as of Sept 2025) and Anthropic’s prompt engineering tutorial (Jupyter notebooks preferred). The checklist highlights concrete, production‑oriented topics and tools: evaluation (Hamel’s evals blog, Hugging Face’s evaluation guidebook and LLM_judge), tracing best practices (run locally for dev, keep tracing off the critical path, prefer open source/affordable options; Langfuse called out), agents (Ampcode/MCP basics and known MCP server issues), and working in production‑grade codebases (Cline, using tools like deepwiki or local repos + AI assistants). An optional stack for deeper model intuition points to 3Blue1Brown and Karpathy LLM/architecture dives. Overall, the list is practical and tooling‑forward: it steers engineers toward robust evals, observability, and code hygiene so deployments scale safely, while flagging resources to deepen LLM fundamentals if you choose.