The Applied AI concept map
214 concepts across 10 tracks, drawn with the 614 links between them. 150 of those links cross tracks, which is the part worth looking at: the concepts that decide interviews are rarely the ones that sit neatly inside one topic. Hover a concept to see only what it touches. Click to read it.
same trackcrosses tracks free membersbigger dot = more connections214 shown
Where the map is dense
The most-connected concepts are the ones the rest of the library keeps reaching for, which makes them the highest-leverage things to be solid on: The RAG Pipeline (18), Embeddings (17), Eval-Driven Development and Golden Datasets (16), The Big-O That Actually Matters (15), and Vector Search and ANN Indexes: HNSW, IVF, Quantization (14). If you are deciding where to spend a week, start with a hub rather than a leaf.
Every concept, by track
Foundations of LLMs & GenAI33
- Embeddings17
- The KV Cache12
- The Transformer Architecture10
- Attention and Self-Attention10
- The Context Window8
- Hallucination8
- Prompting vs RAG vs Fine-Tuning8
- Small vs Large Models and Routing7
- Prompt Engineering6
- Chain-of-Thought and In-Context Learning6
- RLHF: Reinforcement Learning from Human Feedback6
- Reward Models6
- Policy Optimization: PPO and GRPO6
- Inference-Time Compute and Reasoning Models6
- LoRA and Parameter-Efficient Fine-Tuning5
- DPO and Preference-Optimization Variants5
- Training Reasoning Models: RLVR, PRM vs ORM5
- Tokenization4
- Attention Variants: MHA, MQA, and GQA4
- Temperature and Sampling4
- Constrained and Structured Decoding4
- Self-Consistency, Tree-of-Thought, and Prompt Chaining4
- Constitutional AI and RLAIF4
- Scaling Laws4
- Multimodal Models and VLMs4
- Diffusion Models4
- Speech and Voice AI: ASR, TTS, and Voice Agents4
- Context Rot and Long-Context Failure Modes4
- From RNNs to Transformers: RNN, LSTM, Seq2Seq3
- Classic NLP: Bag-of-Words, TF-IDF, and Word2Vec3
- Positional Encodings (RoPE and ALiBi)3
- Mixture-of-Experts3
- Multilingual Models and the Tokenization Tax3
Retrieval & Agents22
- The RAG Pipeline18
- Vector Search and ANN Indexes: HNSW, IVF, Quantization14
- Agents and Tool Use9
- Reranking8
- Hybrid Search and Reciprocal Rank Fusion8
- Multi-Agent Orchestration7
- Retrieval vs Long Context7
- Chunking6
- Agentic and Corrective RAG6
- GraphRAG and Knowledge-Graph Retrieval5
- Query Transformation and Multi-Hop Retrieval5
- Citations and Grounding5
- Function Calling and Tool Schemas5
- Choosing and Adapting Embedding Models4
- Late-Interaction Retrieval (ColBERT)4
- RAPTOR and Small-to-Big: Hierarchical Retrieval for RAG4
- Model Context Protocol (MCP)4
- Agent Memory: Short-Term, Long-Term, and Memory Stores4
- Agent Design Patterns: ReAct, Plan-and-Execute, Reflection4
- Context Engineering for Agents4
- Agent Reliability and Long-Horizon Robustness4
- Agent Evaluation and Trajectory Analysis4
Evaluation & ML Foundations48
- Eval-Driven Development and Golden Datasets16
- Precision, Recall, and F1: Thresholds and Imbalanced Data12
- Calibration and Uncertainty12
- Overfitting and Regularization11
- A/B Testing10
- Offline vs Online Evaluation: Why Offline Wins Fail to Hold9
- Information Theory for ML8
- Probability Distributions: Bernoulli, Normal, and Poisson8
- Cross-Validation (Done Right)8
- RAG Evaluation8
- Gradient Descent and Optimizers7
- CV Architectures: ResNets, ViT, Detection7
- Hypothesis Testing and p-values6
- Linear and Logistic Regression6
- The Bias-Variance Tradeoff6
- Feature Engineering: Encoding, Scaling, Selection6
- Backpropagation, Intuitively6
- Training Neural Nets: Init, Normalization, Dropout, LR Schedules6
- Transfer Learning6
- LLM-as-a-Judge6
- MLE, MAP, and Bayesian vs Frequentist5
- Ensembling: Bagging, Boosting, Stacking5
- Label Noise and Weak Supervision5
- Synthetic Data Generation5
- Activation Functions: ReLU, GELU, SwiGLU5
- Multi-Armed Bandits: Epsilon-Greedy, UCB, Thompson Sampling5
- Benchmarks and Their Limits5
- Contrastive and Metric Learning5
- Catastrophic Forgetting and Continual Learning5
- Object Detection and Segmentation5
- CLT, Sampling, and Confidence Intervals4
- Sampling Techniques: Stratified, Reservoir, Importance4
- Decision Trees and Splitting Criteria4
- SVMs and the Kernel Trick4
- Dimensionality Reduction: PCA, t-SNE, UMAP4
- Imbalanced Data and Resampling4
- Outlier and Anomaly Detection4
- Hyperparameter Optimization4
- Vanishing and Exploding Gradients4
- CNNs: Convolution, Pooling, Receptive Fields4
- Autoencoders and GANs4
- The Computer Vision Pipeline4
- Causal Inference: Confounders and Identification3
- kNN and the Curse of Dimensionality3
- Generative vs Discriminative Models (Naive Bayes)3
- Clustering: K-Means, Hierarchical, DBSCAN3
- Gaussian Mixtures and the EM Algorithm3
- Handling Missing and Corrupted Data3
System Design for AI in Production23
- The LLM Gateway11
- Latency Budgets and Streaming8
- Fault Tolerance and Graceful Degradation8
- Rate Limiting, Retries, and Backoff7
- Idempotency and Exactly-Once Effects7
- LLM Cost Optimization7
- Prompt and Semantic Caching7
- Distributed Key-Value Stores7
- Caching Strategies7
- Guardrails6
- Foundation Model Selection and Benchmarking6
- Observability for LLM Systems5
- Recommendation Systems: Candidate Generation and Ranking5
- Consistent Hashing and Sharding: Hash Ring, Virtual Nodes5
- Load Balancing5
- Prompt Versioning and Management4
- User Feedback Loops and the Data Flywheel4
- Listwise, Pairwise, Pointwise: Learning to Rank Explained4
- Multi-Stage Retrieval and Ranking Funnels4
- CAP and Consistency Models4
- Concurrency and Thread Safety4
- P2P Content Distribution: BitTorrent, Gossip, and CDNs4
- Message Queues and Event Streaming4
ML Infrastructure & Serving12
- GPU Memory and the Serving Stack13
- Quantization and Low Precision8
- Knowledge Distillation8
- Continuous Batching7
- PagedAttention7
- Distributed Training: Parallelism and FSDP5
- Mixed-Precision Training5
- FlashAttention and IO-Aware Kernels4
- Disaggregated Prefill/Decode and Prefix Caching4
- Speculative Decoding4
- Multi-LoRA Serving4
- Model Serving Frameworks4
Data & SQL Engineering23
- Data Quality and Contracts13
- Change Data Capture13
- Idempotent Data Pipelines: Reruns Without Duplicate Rows10
- Incremental Models and MERGE/UPSERT9
- Window Functions8
- Deduplication (Exact and Fuzzy)8
- SQL Joins8
- Gaps and Islands (Sessionization)6
- Query Execution and Optimization5
- Indexing Strategies in SQL: B-Tree, Composite, and Covering5
- Partitioning and Clustering5
- Dimensional Modeling and Star Schemas5
- Backfills and Reprocessing5
- Transactions, ACID, and Isolation Levels4
- GROUP BY and Aggregation4
- CTEs and Subqueries4
- NULLs and Three-Valued Logic4
- Ranking and Top-N Per Group4
- Slowly Changing Dimensions (SCD)4
- Batch vs Streaming4
- Warehouse vs Lake vs Lakehouse4
- Pipeline Orchestration and DAGs4
- Schema Evolution and Data Contracts4
AI Security, Privacy & Governance13
- Prompt Injection8
- Agent Guardrails8
- AI Governance Frameworks7
- PII Handling5
- Audit Trails5
- Multi-Tenancy and Isolation5
- Fairness, Bias, and Model Cards5
- Jailbreaks and Red-Teaming Taxonomy5
- Indirect Prompt Injection and the Lethal Trifecta4
- Differential Privacy4
- Agent Security: Tool Poisoning, Memory Poisoning, Containment4
- Federated Learning3
- Mechanistic Interpretability: Features, Circuits, and SAEs3
Coding & Engineering Craft28
- The Big-O That Actually Matters15
- Arrays and Hashing10
- Two Pointers and Sliding Window: O(n) Array Patterns8
- Trees, BSTs, and Traversal7
- Dynamic Programming6
- Parsing Messy Real-World Data: Defensive Parsing Patterns5
- Streaming and Backpressure5
- Heaps and Priority Queues5
- Graphs: BFS, DFS, and Shortest Paths5
- Recursion and Divide-and-Conquer5
- Greedy Algorithms5
- Sorting Algorithms5
- Linked Lists: Dummy Head, Fast/Slow Pointers, Reversal4
- Stacks and Queues4
- Topological Sort and DAGs4
- Merge Intervals, Meeting Rooms, and the Sweep Line Pattern4
- Tries and String Algorithms4
- Testable Design for AI Systems3
- Binary Search: Off-by-One Templates and Search on the Answer3
- Union-Find (Disjoint Set Union)3
- Backtracking3
- Bit Manipulation3
- Implementing ML From Scratch (NumPy Patterns)3
- Numerical Stability in Code3
- Fast and Slow Pointers (Floyd's Cycle Detection)3
- Monotonic Stack and Queue: Next Greater Element in O(n)3
- Prefix Sums and Difference Arrays3
- Matrix and Grid Simulation Patterns3
The map is for orientation. If you would rather be told what to do in order, the start-here path sequences the same material by background and stage, and the courses walk it front to back.