Skip to content
AI Help Wiki

Project North Star & Scope Policy

Like game database sites (e.g., Fextralife, ESO-Hub) that organize dense tables of weapons, builds, and level walkthroughs, this wiki is a structured database and walkthrough catalog for AI Engineering.

Our North Star is to provide developers with factual, highly structured, and actionable data to help them construct AI “builds” (applications) with optimal components.

In-Scope: The AI Engineering Catalog

Section titled “In-Scope: The AI Engineering Catalog”

Our scope is intentionally broad, focusing on the practical “game mechanics” of AI. The following categories are in-scope:

1. The Component Database (Equivalent to Item databases)

Section titled “1. The Component Database (Equivalent to Item databases)”
  • Model Catalogs: Detailed technical specifications (context windows, tokens-per-second, pricing, capability scores).
  • Tooling Registries: Vector databases (Qdrant, Pinecone, pgvector), agent frameworks (LangGraph, AutoGen), and embedding models (Cohere, OpenAI, Voyage).

2. Implementation “Builds” (Equivalent to Character Build guides)

Section titled “2. Implementation “Builds” (Equivalent to Character Build guides)”
  • Standard recipes for assembling AI applications. For example:
    • The Production RAG Build: Hybrid Search + Reranker + LLM.
    • The Autonomous Agent Build: LangGraph + Tool Calling + Reflection Loop.

3. Step-by-Step Walkthroughs (Equivalent to Quest walkthroughs)

Section titled “3. Step-by-Step Walkthroughs (Equivalent to Quest walkthroughs)”
  • Actionable guides to achieve specific engineering goals:
    • How to fine-tune a model using Unsloth.
    • How to set up local LLM inference with Ollama.
    • How to deploy models to serverless platforms.

4. Mechanics & Formulas (Equivalent to Game scaling formulas)

Section titled “4. Mechanics & Formulas (Equivalent to Game scaling formulas)”
  • Explanations of core mathematical and logical formulas:
    • Tokenization math and cost calculations.
    • Attention scaling limits.
    • Latency vs. throughput formulas.

Out-of-Scope: What We Do Not Accept

Section titled “Out-of-Scope: What We Do Not Accept”

To maintain data integrity and prevent bloat, the following are strictly out-of-scope:

  • AI Corporate Drama & News: Funding announcements, executive hires, or community gossip.
  • Subjective Speculation: Opinion pieces about AGI timelines, societal impacts, or non-technical philosophies.
  • Marketing & Hype: Vague press releases or commercial tool promotions that do not supply verifiable benchmarks, technical architecture maps, or open code repositories.