Gradient · Phase 1

Zero to hero, one chapter at a time.

Machine learning, deep learning, explainable AI, and multimodal AI — every formula visualized, every chapter tested.

Pass a checkpoint to start a streak
Part I — Calculus, Optimization & Gradients10 chapters
Part II — Linear Algebra & Matrix Decompositions10 chapters
Part III — Probability, Information Theory & Bayesian Inference10 chapters
Part IV — Supervised Learning: Regression & Linear Classifiers12 chapters
Part V — Non-Linear Models, Trees, Ensembles & Kernel Methods10 chapters
Part VI — Unsupervised Learning, Clustering, Dimensionality & Time Series11 chapters
Part VII — Model Evaluation, Validation & Feature Engineering10 chapters
Part VIII — Neural Network Fundamentals, Backpropagation & Optimizers10 chapters
Part IX — Deep Learning Regularization, Normalization & Training Dynamics10 chapters
Part X — Computer Vision: CNNs, ResNets, Object Detection & Segmentation11 chapters
Part XI — Sequence Models: RNNs, LSTMs, Attention & The Transformer Block10 chapters
Part XII — Modern Sequence Architectures: RoPE, FlashAttention & State-Space Models10 chapters
Part XIII — Generative Models: VAEs, Flow Matching, Score Models & Diffusion Transformers10 chapters
Part XIV — 3D Vision, Neural Fields & Gaussian Splatting8 chapters
Part XV — Multimodal Foundation Models: CLIP, Cross-Attention & VLMs11 chapters
Part XVI — Reinforcement Learning: Bandits, MDPs, Policy Gradients & PPO10 chapters
Part XVII — LLM Post-Training: SFT, DPO, GRPO & Reasoning / Test-Time Compute11 chapters
Part XVIII — LLM Agents, Tool Use, Planning & Multi-Agent Swarms10 chapters
Part XIX — Alignment, Mechanistic Interpretability, Safety & Red-Teaming10 chapters
Part XX — Embodied AI & Production Systems: VLA Robotics, High-Throughput Serving & MLOps12 chapters
Chapter 1

Embodied AI & Vision-Language-Action (VLA) models

Map visual camera observations directly to low-level robotic motor torques (RT-2/OpenVLA/pi0).

Chapter 2

Diffusion policies for robotic manipulation

Predict smooth, multi-modal continuous physical action trajectories using flow matching.

Chapter 3

Sim-to-real transfer & physics engines (MuJoCo / Isaac)

Bridge the reality gap with domain randomization and high-throughput GPU physics.

Chapter 4

GPU memory hierarchies & kernel profiling

Understand SRAM vs. HBM bandwidth, Tensor Core utilization, and arithmetic intensity.

Chapter 5

Quantization (AWQ, GPTQ, GGUF & BitNet 1.58-bit)

Compress model weights into FP8, INT4, and ternary 1.58-bit representations.

Chapter 6

High-throughput serving: PagedAttention & vLLM

Manage dynamic KV-cache memory blocks without fragmentation to achieve 10x throughput.

Chapter 7

Speculative decoding & Medusa heads

Draft multiple future tokens concurrently and verify in parallel to reduce latency.

Chapter 8

Large-scale distributed training (DDP, FSDP, ZeRO & Megatron-LM)

Distribute massive models across clusters using Tensor, Pipeline, and Sequence Parallelism.

Chapter 9

Vector databases & ANN search (HNSW & IVF-PQ)

Index millions of embeddings and query nearest neighbors in sub-millisecond time.

Chapter 10

Data pipelines & feature stores (Feast)

Ensure point-in-time correctness and synchronize offline training with online serving.

Chapter 11

Model monitoring, drift detection & canary rollouts

Detect covariate and concept drift with Kolmogorov-Smirnov and PSI statistics.

Chapter 12

Build and deploy a monitored, high-throughput AI serving cluster

Deploy a vLLM serving pipeline with INT4 AWQ quantization, Prometheus metrics, and drift alarms.

Part XXI — Explainable AI & Model Interpretability13 chapters
Part XXII — Graph Neural Networks & Structured Data5 chapters
Part XXIII — Modern Architectures, Generative Models & LLM Engineering10 chapters