Skip to content

Contributing

Thanks for helping keep this guide accurate and useful. It is a living reference for production AI systems and interview prep, so contributions grounded in real-world practice are especially welcome.

What to contribute

High-value additions include:

  • Interview questions with staff-level depth and worked answers.
  • Production failure modes you have hit, plus how you diagnosed and fixed them.
  • Model benchmarks and pricing updates for confirmed, publicly released models.
  • Corrections to anything outdated, ambiguous, or wrong.
  • New patterns (RAG, agents, MCP, eval pipelines, multi-tenant isolation) with concrete tradeoffs.

How to propose changes

  1. Fork the repo and create a branch from the latest main.
  2. Make your change in focused commits with logical boundaries.
  3. Open a pull request describing what changed and why. Link a source for any new model, price, or benchmark claim.

For small fixes (typos, broken links, stale numbers), a direct PR is perfect. For larger additions such as a new chapter or case study, open an issue first so we can align on scope.

Style conventions

  • No em dashes. Use commas, parentheses, or a rewrite instead. This applies to prose, commit messages, and code comments.
  • Confirmed-real models only. Reference models that have actually shipped publicly. Do not invent model names, versions, prices, or benchmark numbers. When you cite pricing or capabilities, link the source.
  • Conventional commits. Use messages like docs: fix broken links in case studies or feat(retrieval): add ColBERT reranking notes, scoped to one logical change each.
  • Match the surrounding voice. Opinionated, concrete, and tradeoff-aware. Prefer specific numbers and named tools over vendor-neutral hedging.
  • Keep links relative and verify they resolve before opening a PR.

Quality bar

  • Be accurate. If you are unsure, say so or leave it out.
  • Be concrete. Name the model, the metric, the failure mode, the cost.
  • Be current. This guide tracks what ships, so flag anything that has aged out.

By contributing, you agree that your contributions are licensed under the MIT License.