QuickDevBase

AI knowledge, at a glance

Learn the model.
Ground the answer.
Bound the action.

AI is too broad for one giant checklist. QuickDevBase separates the mental map into three focused paths so each one stays quick, searchable, and genuinely useful.

Start with Generative AI ↗ See the learning order ↓
3Independent paths
36Focused modules
252Quick concepts
01GGenerateModels · prompts · evals
02RGroundRetrieve · rank · cite
03AActTools · control · trace
One account
three certificates

Recommended sequence

Three layers.
One clear progression.

  1. 01
    First, understand generation

    Learn how models, tokens, context, prompts, tools, multimodality, evaluation, and safety fit together.

  2. 02
    Then, ground it with RAG

    Give generation current, private, and citable evidence through ingestion, search, reranking, and evaluation.

  3. 03
    Finally, add bounded action

    Let agents choose tools and coordinate work—with permissions, approvals, traces, limits, and guardrails.

Choose a path

Go wide quickly.
Then go deep officially.

0%ALL PATHS

36 modules across the AI collection

Each path tracks progress and awards its own public completion certificate.
PATH01
G

THE MODEL LAYER

Generative AI
Foundations

The vocabulary behind modern models—from transformers and embeddings to prompts, structured outputs, customization, evals, and production safety.

TransformersTokensPromptingMultimodal
12 modules · 84 conceptsStart here ↗
PATH02
R

THE KNOWLEDGE LAYER

RAG
Systems

The complete evidence pipeline—from parsing and chunking to vector and hybrid search, query rewriting, reranking, citations, evals, and secure operations.

ChunkingEmbeddingsRetrievalCitations
12 modules · 84 conceptsExplore path ↗
PATH03
A

THE ACTION LAYER

Agentic
AI

The systems view of agents—from loops, tools, memory, planning, and orchestration to MCP, human approval, reliability, traces, evaluations, and security.

ToolsMemoryHandoffsGuardrails
12 modules · 84 conceptsExplore path ↗

QuickDevBase is your map—not the source of truth.

Every module ends with a direct primary-documentation link. AI tools, models, limits, and best practices change quickly, so use these lessons to orient yourself and the official sources to implement.
Begin path 01 ↗