Complete Generative AI Masterclass

Topics Covered

• What is Artificial Intelligence? History and evolution
• Types of AI: Narrow AI, General AI, Superintelligence
• Machine Learning vs Deep Learning vs Generative AI
• Generative AI revolution: GPT, current image APIs, Stable Diffusion
• AI career paths and opportunities in 2026 and beyond
• Why Python for AI? Python ecosystem overview
• Installing Python (3.10+) and setting up environment
• Development environments: Jupyter Notebook, VS Code, Google Colab
• Python basics: variables, data types, operators
• Control structures: if-else, loops (for, while)

• Functions: definition, parameters, return values
• Python data structures: lists, tuples, dictionaries, sets
• List comprehensions and dictionary comprehensions
• File handling: reading and writing files
• Error handling: try-except blocks
• Python modules and packages
• Virtual environments with venv/conda

Topics Covered

• Object-Oriented Programming in Python
• Classes and objects: __init__, self
• Inheritance and polymorphism
• Magic methods: __str__, __repr__, __len__
• Decorators and their usage
• Lambda functions and functional programming
• Map, filter, reduce functions
• Generators and iterators
• Context managers and 'with' statement
• Regular expressions for text processing

• Working with JSON and CSV data
• Introduction to NumPy: arrays and operations
• NumPy array operations, broadcasting, indexing
• Introduction to Pandas: DataFrames and Series
• Data loading, cleaning, and exploration with Pandas
• Data visualization basics with Matplotlib
• Plotting: line plots, bar charts, scatter plots, histograms

Topics Covered

• Why math matters for AI and ML
• Linear Algebra: vectors and vector operations
• Dot product and cross product
• Matrices and matrix operations
• Matrix multiplication and transposition
• Identity matrix and inverse matrix
• Eigenvalues and eigenvectors (intuition)
• Calculus: functions and limits
• Derivatives: concepts and rules
• Chain rule and partial derivatives

• Gradient: direction of steepest ascent
• Gradient descent algorithm basics
• Multivariable calculus for neural networks
• Probability theory: basics and terminology
• Probability distributions: uniform, normal (Gaussian)
• Expected value and variance
• Bayes' theorem and conditional probability

Topics Covered

• Statistics for data science
• Descriptive statistics: mean, median, mode
• Variance and standard deviation
• Correlation and covariance
• Hypothesis testing basics
• P-values and statistical significance
• Central Limit Theorem
• Confidence intervals
• Linear regression: mathematical foundation
• Cost functions and loss functions

• Optimization: local minima, global minima
• Convex vs non-convex functions
• Introduction to information theory
• Entropy and cross-entropy
• KL divergence
• Dimensionality reduction concepts
• Principal Component Analysis (PCA) intuition

Topics Covered

• What is Machine Learning? Formal definition
• Types of learning: Supervised, Unsupervised, Reinforcement
• Machine learning workflow: data, model, evaluation
• Training, validation, and test sets
• Overfitting and underfitting
• Bias-variance tradeoff
• Introduction to scikit-learn library
• Linear regression with scikit-learn
• Logistic regression for classification
• Decision trees: intuition and implementation

• Random forests ensemble method
• K-Nearest Neighbors (KNN) algorithm
• Support Vector Machines (SVM) basics
• Naive Bayes classifier
• Model evaluation metrics: accuracy, precision, recall, F1
• Confusion matrix and ROC curves
• Cross-validation techniques

Topics Covered

• From ML to Deep Learning: why neural networks?
• Perceptron: the building block
• Multi-layer perceptron (MLP) architecture
• Activation functions: sigmoid, tanh, ReLU, Leaky ReLU
• Forward propagation step-by-step
• Loss functions: MSE, cross-entropy
• Backpropagation algorithm intuition
• Gradient descent and its variants (SGD, Adam)
• Learning rate and hyperparameters
• Introduction to TensorFlow and Keras

• Building neural networks with Keras Sequential API
• Training and evaluating neural networks
• Regularization techniques: dropout, L1/L2
• Batch normalization
• Early stopping and model checkpoints
• Introduction to PyTorch basics
• PyTorch tensors and autograd
• What generative means: from predicting labels to producing text, images and code

Topics Covered

• Limitations of RNNs and LSTMs
• Attention mechanism: intuition and math
• Self-attention and multi-head attention
• Positional encoding in transformers
• Transformer architecture: encoder-decoder
• BERT: bidirectional encoder representations
• GPT architecture: decoder-only transformers
• Tokenization: BPE, WordPiece, SentencePiece
• Embedding layers and embedding spaces
• How GPT generates text: autoregressive generation

• Temperature, top-k, top-p sampling
• Model parameters: understanding 7B, 70B, 175B
• Context windows and attention span
• Fine-tuning vs prompt engineering vs RAG
• Pre-training and transfer learning
• Hugging Face Transformers library introduction
• Loading and using pre-trained models

Topics Covered

• OpenAI API overview and capabilities
• Getting API keys and authentication
• Current OpenAI chat models, and how to choose between them
• Chat completions API: messages and roles
• System, user, and assistant messages
• Completion parameters: temperature, max_tokens, top_p
• Streaming responses for real-time output
• Function calling with frontier models
• JSON mode for structured outputs
• Vision API: frontier modelsV for image understanding

• current image APIs API for image generation
• Whisper API for speech-to-text
• TTS (Text-to-Speech) API
• Embeddings API and semantic search
• Moderation API for content filtering
• Token counting and cost optimization
• Rate limiting and error handling
• Claude and Gemini APIs, and the OpenAI-compatible pattern that unifies the field

Topics Covered

• Prompt engineering as a discipline
• Anatomy of effective prompts
• Zero-shot, one-shot, few-shot learning
• Chain-of-Thought (CoT) prompting
• Tree of Thoughts (ToT) technique
• Self-consistency prompting
• Role prompting and persona design
• Instruction tuning concepts
• Prompt templates and variables
• Output formatting and constraints

• Handling hallucinations and reducing errors
• Prompt injection and safety
• Meta-prompting and prompt generation
• Multi-turn conversation design
• Context management in long conversations
• Prompt optimization and A/B testing
• Domain-specific prompting strategies

Topics Covered

• What are embeddings? Semantic representations
• Word2Vec, GloVe, and modern embeddings
• OpenAI embeddings (text-embedding-ada-002)
• Sentence embeddings and document embeddings
• Cosine similarity and semantic search
• Vector databases: purpose and architecture
• Pinecone: managed vector database
• Weaviate: open-source vector DB
• Chroma: simple vector store
• Qdrant: performance-focused vector DB

• FAISS: Facebook AI similarity search
• Indexing strategies: flat, HNSW, IVF
• Metadata filtering in vector search
• Hybrid search: vector + keyword
• Building semantic search systems
• Recommendation systems with embeddings
• Clustering and visualization of embeddings

Topics Covered

• Introduction to LangChain framework
• LangChain architecture and components
• LLMs and Chat Models in LangChain
• Prompt templates and prompt engineering
• Output parsers: structured data from LLMs
• Chains: sequential processing
• LLMChain, SimpleSequentialChain, SequentialChain
• Memory types: ConversationBufferMemory, ConversationSummaryMemory
• Conversation chains with memory
• Document loaders: PDF, CSV, web pages

• Text splitters and chunking strategies
• Vector stores integration in LangChain
• Retrievers for information retrieval
• Combining LLMs with external data
• Question-answering over documents
• LangChain Expression Language (LCEL)
• Debugging and tracing in LangChain
• Framework honesty: when LangChain-style tooling helps, and when the native SDK is cleaner

Topics Covered

• What is RAG? Why RAG over fine-tuning?
• RAG architecture: retrieval + generation
• Building basic RAG pipeline
• Document preprocessing for RAG
• Chunking strategies: fixed-size, semantic
• Embedding generation for documents
• Vector store population
• Retrieval strategies: similarity search, MMR
• Reranking retrieved documents
• Context compression techniques

• Query transformation and expansion
• Retrieval quality evaluation
• Prompt engineering for RAG
• Handling long contexts
• Multi-query retrieval
• Hypothetical document embeddings (HyDE)
• Parent document retrieval

Topics Covered

• Advanced RAG architectures
• Self-RAG: self-reflective retrieval
• Corrective RAG (CRAG)
• Agentic RAG with LangGraph
• Multi-document RAG systems
• Hierarchical retrieval strategies
• Graph-based RAG
• Conversational RAG with memory
• RAG with structured data (tables, charts)
• Multimodal RAG: text + images

• RAG evaluation metrics: faithfulness, relevance
• RAGAS framework for evaluation
• RAG optimization techniques
• Caching strategies for RAG
• Production RAG deployment
• RAG monitoring and observability
• Security and privacy in RAG systems

Topics Covered

• What are AI agents? Autonomous systems
• Agent types: simple reflex, goal-based, utility-based
• Perceive-Think-Act loop
• ReAct: reasoning and acting
• Plan-and-Execute agents
• Cognitive architectures for agents
• Agent memory systems
• Short-term vs long-term memory
• Semantic memory and episodic memory
• Working memory implementation

• Tool use and function calling
• Multi-step reasoning with agents
• Error handling and recovery
• Agent observability and logging
• LangGraph for agent workflows
• State machines for agents
• Human-in-the-loop agents

Topics Covered

• Multi-agent systems (MAS)
• Agent communication protocols
• Collaborative agents vs competitive agents
• Agent coordination and cooperation
• Hierarchical agent architectures
• AutoGPT and BabyAGI patterns
• Task decomposition and planning
• Sub-goal generation
• Agent self-evaluation and reflection
• MetaGPT and role-playing agents

• Specialized agent roles
• Agent orchestration frameworks
• Crew AI for multi-agent systems
• Agent security and sandboxing
• Rate limiting and cost control
• Agent performance optimization
• Debugging complex agent systems

Topics Covered

• What is fine-tuning? When to fine-tune?
• Fine-tuning vs prompt engineering vs RAG
• Types of fine-tuning: full, parameter-efficient
• Transfer learning for LLMs
• Dataset preparation for fine-tuning
• Data annotation and labeling
• Instruction tuning datasets
• Supervised fine-tuning (SFT)
• Training loops and optimization
• Loss functions for language models

• Hyperparameter tuning
• Overfitting prevention in fine-tuning
• Evaluation metrics for fine-tuned models
• Benchmark datasets and tasks
• OpenAI fine-tuning API
• Fine-tuning current OpenAI models
• Cost considerations for fine-tuning

Topics Covered

• Open source LLM landscape
• Llama 2 and Llama 3 models
• Mistral and Mixtral models
• Falcon, MPT, and BLOOM
• Model cards and model documentation
• Downloading models from Hugging Face Hub
• Model quantization: GGUF, GPTQ, AWQ
• Running LLMs locally with Ollama
• llama.cpp for efficient inference
• vLLM for production serving

• Text Generation Inference (TGI)
• Model serving with FastAPI
• GPU vs CPU inference
• Batch inference optimization
• Caching and KV-cache optimization
• Model benchmarking: speed and quality
• Building custom inference servers
• Multimodal models in one honest lesson: vision-language and where it fits

Topics Covered

• Designing AI-powered applications
• Microservices architecture for AI
• FastAPI for ML/AI APIs
• Request validation and error handling
• Asynchronous processing with Celery
• Message queues: RabbitMQ, Redis
• Background tasks for AI workloads
• Caching strategies for AI apps
• Redis for response caching
• Database integration: PostgreSQL, MongoDB

• Storing embeddings in databases
• User authentication and authorization
• API rate limiting and throttling
• Webhook integrations
• Building SDKs for AI APIs
• API documentation with OpenAPI
• Postman and API testing
• LLM observability: tracing, token accounting and quality dashboards
• Security: prompt injection, data exfiltration and the guardrail stack

Topics Covered

• AI ethics fundamentals
• Bias in AI: sources and mitigation
• Fairness metrics and evaluation
• Transparency and explainability
• LIME and SHAP for model interpretation
• Privacy in AI: data protection
• Differential privacy
• Federated learning basics
• AI safety and alignment
• Constitutional AI principles

• Red teaming and adversarial testing
• Content moderation and filtering
• Deepfakes: detection and prevention
• AI regulations: GDPR, AI Act
• Responsible AI deployment
• Environmental impact of AI
• AI for social good
• Capstone planning: choosing a system worth defending

Topics Covered

• Two focused weeks: a complete GenAI system, RAG, agent or hybrid
• Instructor checkpoints and honest scope cuts
• The evaluation harness built before the excitement
• Shipped behind an API with observability and guardrails
• A system card a skeptic could audit

Topics Covered

• Presenting the capstone: live system plus the hardest failure story
• A full-course spiral review before the exam
• The honest map of next roads: enterprise agents, coding agents, deep ML and research
• Where our Copilot Studio and Codex/Claude Code courses continue the applied layer