fig_002 · curriculum mindmap
The Roadmap 20 phases, 435 lessons. Phases are arranged in tiers — anything on the same row can be learned in parallel. Pick a phase and flip on dependency mode to see every prerequisite light up.
show dependenciesfocus → phase 3
prerequisite on the path same tier · parallel okstart here
Aye Intelligence
435 lessons · 20 phases
tier 0
0 phase 0
Setup & Tooling warm up the workshop
+ tier 1
I phase 1 · needs 0
Math Foundations vectors, calculus, probability
+ tier 2· parallel
II phase 2 · needs 1
ML Fundamentals regression to ensembles
+ III phase 3 · needs 1, 2
Deep Learning Core backprop by hand
− tier 3· parallel
IV phase 4 · needs 3
Computer Vision CNNs, ViTs, segmentation
+ V phase 5 · needs 3
NLP Foundations → Advanced tokens to transformers
+ VI phase 6 · needs 3
Speech & Audio spectrograms, ASR, TTS
+ IX phase 9 · needs 3
Reinforcement Learning policy & value methods
+ tier 4
VII phase 7 · needs 3, 5
Transformers Deep Dive attention, KV cache
+ tier 5· parallel
VIII phase 8 · needs 7, 4
Generative AI diffusion, VAEs, GANs
+ X phase 10 · needs 7
LLMs from Scratch build a tiny GPT
+ XII phase 12 · needs 7, 4, 5
Multimodal AI vision + language + audio
+ tier 6
XI phase 11 · needs 10
LLM Engineering prompts, RAG, evals
+ tier 7· parallel
XIII phase 13 · needs 11
Tools & Protocols MCP, function calling
+ XVII phase 17 · needs 10, 11
Infra & Production serving, scaling, ops
+ tier 8
XIV phase 14 · needs 11, 13
Agent Engineering loops, memory, planning
+ tier 9· parallel
XV phase 15 · needs 14
Autonomous Systems long-running agents
+ XVI phase 16 · needs 14
Multi-Agent & Swarms cooperation & roles
+ XVIII phase 18 · needs 11, 14
Ethics, Safety & Alignment the human side
+ tier 10
XIX phase 19 · needs 11, 14, 17
Capstone Projects ship something real
+