English · Español

Study any chapter

The ideal path is sequential — the dependency chain in LYNX_CORTEX.md §6 exists for a reason. But the curriculum also works as a reference work: every row links to the chapter, lists its direct conceptual prerequisites (informational, never locking), the phases it unlocks, and the concepts it teaches. Use site search or this table to jump straight to the topic you need.

Dependency graph

graph LR
  P00["00 Project Foundations & Learning Methodology"]
  P01["01 Hardware & Computing Substrate"]
  P02["02 Numerical Representation"]
  P03["03 Linear Algebra from First Principles"]
  P04["04 Calculus & Optimization for AI"]
  P05["05 Probability & Information Theory"]
  P06["06 Python for AI Engineering"]
  P07["07 Scalar Autograd from Scratch"]
  P08["08 Tensor Autograd from Scratch"]
  P09["09 Tiny MLP & Module Abstraction"]
  P10["10 Initialization, Normalization, Residuals"]
  P11["11 Tokenization Theory + BPE Implementation"]
  P12["12 The Corpus"]
  P13["13 Embeddings & Representation Spaces"]
  P14["14 Pre-Transformer Sequence Models"]
  P15["15 Attention from Scratch"]
  P16["16 Positional Encodings"]
  P17["17 Tiny Transformer Block & Mini-GPT"]
  P18["18 Training Loop, Mixed Precision Preview, Checkpointing"]
  P19["19 Training Dynamics & Debugging"]
  P20["20 Evaluation Harness"]
  P21["21 Inference Internals & Sampling"]
  P22["22 KV Cache"]
  P23["23 GPU Architecture Fundamentals"]
  P24["24 CUDA & Triton Hands-On"]
  P25["25 PyTorch Internals"]
  P26["26 Quantization Deep Dive"]
  P27["27 Modern Attention Optimizations"]
  P28["28 Fine-Tuning, LoRA, QLoRA"]
  P29["29 Retrieval-Augmented Generation"]
  P30["30 Structured Generation & Constrained Decoding"]
  P31["31 Tool Use & the Model Context Protocol"]
  P32["32 Agents"]
  P33["33 Inference Serving"]
  P34["34 Observability, Cost & Capacity"]
  P35["35 Distributed Training & Inference"]
  P36["36 Frontier Architectures"]
  P37["37 Security & Safety of AI Systems"]
  P38["38 Cost, Capacity, Operations, MLOps"]
  P39["39 Capstone"]
  P40["40 Hardening, Postmortem, 'What's Next'"]
  P41["41 Learner Portal"]
  PX1["X1 Pretraining at Scale"]
  PX2["X2 Multi-Modal Models"]
  PX3["X3 RLHF / DPO / RLAIF"]
  PX4["X4 Hardware Deep-Dive"]
  PX5["X5 Interview Prep"]
  P00 --> P01
  P01 --> P02
  P02 --> P03
  P03 --> P04
  P04 --> P05
  P05 --> P06
  P04 --> P07
  P06 --> P07
  P07 --> P08
  P08 --> P09
  P09 --> P10
  P10 --> P11
  P11 --> P12
  P11 --> P13
  P12 --> P13
  P13 --> P14
  P13 --> P15
  P14 --> P15
  P15 --> P16
  P10 --> P17
  P15 --> P17
  P16 --> P17
  P12 --> P18
  P17 --> P18
  P18 --> P19
  P19 --> P20
  P20 --> P21
  P15 --> P22
  P21 --> P22
  P01 --> P23
  P22 --> P23
  P23 --> P24
  P08 --> P25
  P24 --> P25
  P02 --> P26
  P25 --> P26
  P15 --> P27
  P22 --> P27
  P26 --> P27
  P26 --> P28
  P27 --> P28
  P13 --> P29
  P28 --> P29
  P21 --> P30
  P29 --> P30
  P30 --> P31
  P29 --> P32
  P31 --> P32
  P22 --> P33
  P32 --> P33
  P33 --> P34
  P18 --> P35
  P34 --> P35
  P35 --> P36
  P29 --> P37
  P31 --> P37
  P36 --> P37
  P37 --> P38
  P38 --> P39
  P39 --> P40
  P40 --> P41
  P35 --> PX1
  P15 --> PX2
  P17 --> PX2
  P28 --> PX2
  P05 --> PX3
  P19 --> PX3
  P28 --> PX3
  P23 --> PX4
  P35 --> PX4
  P40 --> PX5

Phase index

Phase	Chapter	Requires	Unlocks	Teaches
00	Project Foundations & Learning Methodology	—	01	reproducibility seeding lockfile pre-commit ci mypy
01	Hardware & Computing Substrate	00	02 23	memory-hierarchy roofline arithmetic-intensity cache latency bandwidth
02	Numerical Representation	01	03 26	ieee-754 floating-point softmax-stability log-sum-exp precision denormals
03	Linear Algebra from First Principles	02	04	tensors matmul einsum svd rank norms
04	Calculus & Optimization for AI	03	05 07	gradients chain-rule backprop sgd momentum adam
05	Probability & Information Theory	04	06 X3	probability entropy kl-divergence cross-entropy mle perplexity
06	Python for AI Engineering	05	07	numpy strides broadcasting views vectorization profiling
07	Scalar Autograd from Scratch (minigrad)	04 06	08	autograd computation-graph reverse-mode topological-sort dag
08	Tensor Autograd from Scratch	07	09 25	tensor-autograd broadcasting-backward gradcheck matmul-grad softmax-grad
09	Tiny MLP & Module Abstraction (minitorch)	08	10	module-abstraction parameter-registration linear sequential optimizers
10	Initialization, Normalization, Residuals	09	11 17	initialization xavier kaiming layer-norm rms-norm residuals pre-ln
11	Tokenization Theory + BPE Implementation	10	12 13	tokenization bpe byte-level subwords vocabulary zipf
12	The Corpus: Designing the Microscopic Dataset	11	13 18	corpus-design enumeration stratified-split data-manifest reproducibility
13	Embeddings & Representation Spaces	11 12	14 15 29	embeddings cbow cosine-similarity dimensionality representation-geometry
14	Pre-Transformer Sequence Models	13	15	n-gram rnn lstm vanishing-gradient perplexity-baselines
15	Attention from Scratch	13 14	16 17 22 27 X2	attention scaled-dot-product multi-head causal-mask query-key-value
16	Positional Encodings	15	17	positional-encoding rope sinusoidal extrapolation
17	Tiny Transformer Block & Mini-GPT	10 15 16	18 X2	transformer-block pre-ln ffn gelu tied-embeddings lm-head
18	Training Loop, Mixed Precision Preview, Checkpointing	12 17	19 35	training-loop batching adamw warmup cosine-decay checkpointing
19	Training Dynamics & Debugging	18	20 X3	instrumentation hooks gradient-norms loss-curves debugging
20	Evaluation Harness	19	21	evaluation accuracy-probes calibration adversarial-eval bootstrap-ci
21	Inference Internals & Sampling	20	22 30	sampling temperature top-k top-p decode-cost
22	KV Cache: From Math to Memory	15 21	23 27 33	kv-cache prefill decode memory-bound arithmetic-intensity
23	GPU Architecture Fundamentals	01 22	24 X4	gpu-architecture sm warps occupancy hbm coalescing
24	CUDA & Triton Hands-On	23	25	cuda triton kernels shared-memory tiling
25	PyTorch Internals	08 24	26	dispatcher autograd-graph torch-compile custom-ops
26	Quantization Deep Dive	02 25	27 28	quantization int8 nf4 gptq gguf calibration
27	Modern Attention Optimizations	15 22 26	28	flash-attention online-softmax paged-attention gqa mqa
28	Fine-Tuning, LoRA, QLoRA	26 27	29 X2 X3	fine-tuning sft lora qlora catastrophic-forgetting
29	Retrieval-Augmented Generation (RAG)	13 28	30 32 37	rag chunking bm25 hnsw hybrid-search reranking
30	Structured Generation & Constrained Decoding	21 29	31	structured-generation json-mode logit-masking grammar-constrained-decoding
31	Tool Use & the Model Context Protocol (MCP)	30	32 37	tool-use function-calling mcp json-schema json-rpc
32	Agents: Planning, Memory, Sandboxing (Grammar Tutor)	29 31	33	agents react-loop planning agent-memory sandboxing
33	Inference Serving: From FastAPI to Continuous Batching	22 32	34	serving continuous-batching scheduling littles-law load-testing
34	Observability, Cost & Capacity	33	35	observability red-metrics opentelemetry prometheus grafana cost-accounting
35	Distributed Training & Inference	18 34	36 X1 X4	distributed-training ddp fsdp tensor-parallel allreduce nccl
36	Frontier Architectures	35	37	mixture-of-experts mamba state-space-models speculative-decoding
37	Security & Safety of AI Systems	29 31 36	38	prompt-injection jailbreaks threat-modeling supply-chain-security red-teaming
38	Cost, Capacity, Operations, MLOps	37	39	mlops model-registry ab-testing canary-deploys drift-detection
39	Capstone: The Miniature Production System	38	40	integration end-to-end-demo architecture-diagrams cost-reporting
40	Hardening, Postmortem, "What's Next"	39	41 X5	postmortem hardening performance-audit synthesis
41	Learner Portal: Delivering the Curriculum	40	—	web-portal argon2id csrf spaced-repetition multi-tenancy
X1	Pretraining at Scale	35	—	scaling-laws chinchilla mfu pretraining cluster-economics
X2	Multi-Modal Models	15 17 28	—	vit clip whisper contrastive-learning multimodal-fusion
X3	RLHF / DPO / RLAIF	05 19 28	—	rlhf reward-modeling ppo dpo constitutional-ai
X4	Hardware Deep-Dive	23 35	—	h100 nvlink allreduce interconnects datacenter-economics
X5	Interview Prep	40	—	whiteboarding systems-design paper-reading storytelling