AI Concepts — Deep Reference

Core AI/ML concepts every developer working with LLMs and AI systems should understand. Organized by category. Each concept includes what it is, how it works internally, code examples, and practical context.

Files

#	File	Topics
1	01-llm-fundamentals.md	Transformer architecture, self-attention, multi-head attention, positional encoding, encoder vs decoder, GPT vs BERT architecture comparison
2	02-tokenization-and-embeddings.md	BPE, SentencePiece, WordPiece, token limits, embedding spaces, word2vec, semantic similarity, dimensionality
3	03-prompt-engineering.md	Zero-shot, few-shot, chain-of-thought, system prompts, temperature/top-p, structured output, prompt injection risks
4	04-fine-tuning-and-training.md	Pre-training vs fine-tuning, LoRA/QLoRA, RLHF, DPO, instruction tuning, dataset preparation, evaluation
5	05-rag-fundamentals.md	RAG pipeline overview, when to use RAG vs fine-tuning, chunking strategies, retrieval + generation flow, hallucination reduction
6	06-vector-databases.md	Vector DB concepts (indexing, ANN search, HNSW, IVF), Pinecone, Chroma, pgvector, Weaviate, comparison and trade-offs
7	07-embeddings-and-similarity.md	Embedding models (OpenAI, Cohere, open-source), cosine similarity, dot product, distance metrics, re-ranking, hybrid search
8	08-mcp-model-context-protocol.md	What MCP is, architecture (host/client/server), transport (stdio, SSE), tools/resources/prompts, building MCP servers, security
9	09-tool-use-and-function-calling.md	Function calling patterns, tool schemas (JSON Schema), orchestration loops, error handling, parallel tool calls
10	10-ai-agents-and-orchestration.md	Agent architectures (ReAct, plan-and-execute), LangChain, LlamaIndex, memory patterns, multi-agent systems
11	11-ml-fundamentals.md	Neural network basics, backpropagation, loss functions, optimizers, overfitting/regularization, batch normalization
12	12-ml-architectures.md	CNNs, RNNs/LSTMs, attention mechanism evolution, diffusion models, GANs, training vs inference trade-offs