Contents¶
Every part and every chapter, on one page. The same tree lives in the left sidebar, where each part expands in place.
Home¶
Contents¶
Preface¶
I. Math¶
- Linear algebra
- Calculus & matrix calculus
- Probability
- Statistics
- Information theory
- Optimization
- Tensors, shapes & broadcasting
II. Classical ML¶
- Linear regression
- Logistic & softmax regression
- Trees & ensembles
- KNN & K-means
- Probabilistic models & EM
- Dimensionality reduction
- Kernel methods & SVMs
III. Neural Nets¶
- MLPs & activations
- Backpropagation
- Building an autograd engine
- Initialization
- Normalization
- Regularization
IV. Vision¶
- Image representation & signal processing
- Convolutions
- CNN architectures
- Object detection
- Segmentation
- Geometry & cameras
- 3D perception
V. Sequences & Transformers¶
- RNN, LSTM, GRU
- Seq2seq & early attention
- Attention mathematics
- Transformer architectures
- Positional encodings
- Tokenization
VI. LLM Training¶
- Pretraining data & objective
- Scaling laws
- Large-model architecture (MoE, GQA, SSMs)
- Efficient attention & KV cache
- Quantization
- Fine-tuning & LoRA
- Mid-training
VII. Post-training¶
- Supervised fine-tuning
- Reward models & preferences
- RLHF with PPO
- DPO and its relatives
- Reasoning RL, RLVR & GRPO
- Test-time compute
VIII. Multimodal¶
- Vision Transformers
- DETR & set prediction
- CLIP & contrastive learning
- VLM architecture
- Multimodal foundation models
- Video models
IX. Generative¶
X. Self-/Semi-/Weak Supervision¶
XI. Perception & Autonomy¶
- Perception foundation models
- Multi-camera & BEV
- Sensor fusion
- Tracking
- Occupancy & temporal perception
- Prediction & planning
- World models
XII. Reinforcement Learning¶
- MDPs & Bellman equations
- Classical RL algorithms
- Deep RL & DQN
- Policy gradients, GAE & PPO
- Imitation learning
- Agents & tool use
XIII. Retrieval, Eval & Reliability¶
XIV. Systems¶
XV. Interpretability & Safety¶
XVI. Coding Canon¶
XIX. Breadth Review¶
- Basics, losses, models
- Evaluation and debugging
- Deep learning and Transformers
- LLMs and multimodal
- Serving, data engines, real-world
- Rapid-fire derivations
XX. Your Projects¶
- The CARL framework
- Worked deep dive, StreetSmart
- Worked deep dive, edge perception
- Behavioral and leadership
XVII. ML System Design¶
- The framework
- Feed & recommendation ranking
- Search ranking
- Ads CTR / conversion prediction
- Fraud & anomaly detection
- Autonomous-vehicle perception
- Visual search & image retrieval
- Content moderation & trust
- LLM assistant with RAG
- OCR & document understanding
- Novel-object 3D detection from unlabeled fleet data
- Forecasting & ETA
- Notifications, uplift & experimentation
- ML platform, feature store & monitoring
XVIII. Company Deep Dives¶
- Tesla (Autopilot / FSD / Optimus)
- Waymo
- Zoox, Nuro, Aurora & robotaxi peers
- NVIDIA (DRIVE, Cosmos, Isaac)
- Meta (Feed, Reels, Ads, FAIR)
- Google & YouTube
- TikTok / ByteDance
- Netflix & Spotify
- Uber & DoorDash
- Airbnb
- Amazon
- Apple
- Stripe & fintech fraud
- OpenAI, Anthropic & DeepMind
- Scale AI & data engines
XXI. Zipline Offboard Perception¶
- Frames and camera geometry
- Two-view geometry and triangulation
- SfM and bundle adjustment
- Learned 3-D and splats
- Deliverability system design
- Evaluation and scaling
- Coding drills
- Mock interview
- Story bank
- Cheatsheet