AI and Machine Learning Masterclass

Topics Covered

• Why mathematics matters in AI/ML
• Basic arithmetic and algebra review
• Functions and graphs
• Introduction to vectors: geometric and algebraic view
• Vector operations: addition, scalar multiplication
• Dot product and geometric interpretation
• Vector norms (L1, L2, infinity norm)
• Matrices: definition and notation
• Matrix operations: addition, multiplication
• Matrix transpose and symmetric matrices

• Identity matrix and inverse matrices
• Matrix properties and rules
• Systems of linear equations
• Gaussian elimination
• Matrix rank and determinant (basics)

Topics Covered

• Eigenvalues and eigenvectors: concept and computation
• Eigendecomposition
• Singular Value Decomposition (SVD)
• Principal Component Analysis (PCA) mathematics
• Vector spaces and subspaces
• Linear transformations
• Introduction to calculus: limits
• Derivatives: definition and rules
• Power rule, product rule, quotient rule, chain rule
• Partial derivatives

• Gradient: vector of partial derivatives
• Gradient descent concept
• Second derivatives and Hessian matrix
• Taylor series expansion
• Optimization basics: finding minima and maxima

Topics Covered

• Probability fundamentals: sample space, events
• Probability rules: addition, multiplication
• Conditional probability and Bayes' theorem
• Independent vs dependent events
• Random variables: discrete and continuous
• Probability distributions: uniform, Bernoulli, binomial
• Normal (Gaussian) distribution: properties and applications
• Poisson distribution
• Exponential distribution
• Probability density functions (PDF)

• Cumulative distribution functions (CDF)
• Expected value (mean) and variance
• Standard deviation and standard error
• Covariance and correlation
• Law of large numbers and Central Limit Theorem

Topics Covered

• Descriptive statistics: mean, median, mode
• Measures of spread: range, IQR, variance, std dev
• Hypothesis testing fundamentals
• Null and alternative hypotheses
• P-values and significance levels
• T-tests, Z-tests, Chi-square tests
• ANOVA (Analysis of Variance)
• Confidence intervals
• Statistical inference
• Python setup for ML: Anaconda, Jupyter
• NumPy for numerical computing
• Pandas for data manipulation
• Matplotlib and Seaborn for visualization
• Statistical computing with SciPy

Topics Covered

• Advanced NumPy: broadcasting, vectorization
• NumPy for linear algebra
• NumPy random number generation
• Pandas DataFrames: advanced operations
• Data cleaning: handling missing values
• Data transformation: apply, map, applymap
• GroupBy operations and aggregations
• Merging, joining, and concatenating
• Time series data handling
• Categorical data handling
• Multi-index DataFrames
• Performance optimization in Pandas

Topics Covered

• Principles of data visualization
• Matplotlib deep dive: customization
• Seaborn for statistical plots
• Plotly for interactive visualizations
• Distribution plots: histograms, KDE, box plots
• Relationship plots: scatter, line, regression
• Categorical plots: bar, count, violin
• Heatmaps and correlation matrices
• Pair plots and joint plots
• Exploratory Data Analysis (EDA) process

• Identifying patterns and anomalies
• Feature correlation analysis
• Handling outliers
• Data storytelling with visualizations
• SQL for data work: the one honest lesson every ML practitioner needs

Topics Covered

• What is Machine Learning? AI vs ML vs DL
• Types of ML: supervised, unsupervised, reinforcement
• ML workflow: problem definition to deployment
• Data preparation for ML
• Train-test split and validation set
• Cross-validation: k-fold, stratified k-fold
• Overfitting and underfitting
• Bias-variance tradeoff
• Evaluation metrics: accuracy, precision, recall, F1
• Confusion matrix

• ROC curve and AUC
• Scikit-learn library introduction
• Linear Regression: theory and mathematics
• Gradient descent for linear regression
• Cost function (MSE) and optimization

Topics Covered

• Logistic Regression: binary classification
• Sigmoid function and probability interpretation
• Log loss (binary cross-entropy)
• Multi-class classification: one-vs-rest, softmax
• Decision Trees: splitting criteria (Gini, entropy)
• Information gain and entropy
• Tree pruning to prevent overfitting
• Random Forest: ensemble of trees
• Bagging and bootstrap aggregating
• Feature importance in Random Forest

• K-Nearest Neighbors (KNN): distance-based learning
• Choosing K and distance metrics
• Naive Bayes classifier: probabilistic approach
• Support Vector Machines (SVM): margin maximization
• Kernel trick for non-linear classification

Topics Covered

• Feature engineering importance
• Feature scaling: standardization vs normalization
• Min-Max scaling and Standard scaling
• Robust scaling for outliers
• Handling categorical features: one-hot encoding
• Label encoding and ordinal encoding
• Target encoding
• Feature creation: polynomial features
• Interaction features
• Binning and discretization

• Text features: Bag of Words, TF-IDF
• Feature selection: filter methods
• Wrapper methods: forward/backward selection
• Embedded methods: Lasso, Ridge
• Dimensionality reduction: PCA

Topics Covered

• Clustering: grouping similar data
• K-Means clustering: algorithm and initialization
• Elbow method for choosing K
• Hierarchical clustering: agglomerative and divisive
• Dendrograms
• DBSCAN: density-based clustering
• Gaussian Mixture Models (GMM)
• Anomaly detection techniques
• Dimensionality reduction: PCA (practical)
• t-SNE for visualization
• UMAP for dimensionality reduction
• Association rule mining: Apriori algorithm
• Market basket analysis
• Autoencoders for unsupervised learning (intro)

Topics Covered

• Evaluation metrics deep dive
• Classification metrics: precision, recall, F1, AUC-ROC
• Multi-class metrics: macro vs micro averaging
• Regression metrics: MAE, MSE, RMSE, R²
• Custom metrics creation
• Cross-validation strategies
• Stratified sampling
• Time series cross-validation
• Model interpretation: SHAP values
• LIME for local interpretability

• Feature importance analysis
• Partial dependence plots
• Model debugging and error analysis
• Bias and fairness in ML models
• A/B testing for ML models
• Ensemble methods: bagging, boosting and when they stop helping

Topics Covered

• Complete ML pipeline development
• Problem definition and data collection
• EDA and feature engineering
• Model selection and training
• Hyperparameter tuning
• Model evaluation and interpretation
• Documentation and presentation

Topics Covered

• Why deep learning?
• Neural networks basics: perceptron
• Activation functions: sigmoid, tanh, ReLU
• Forward propagation
• Backpropagation algorithm
• Gradient descent variants: SGD, momentum, Adam
• Loss functions for deep learning
• TensorFlow basics
• Keras Sequential API
• Building first neural network
• Training deep networks
• Regularization: dropout, L1, L2
• Batch normalization

Topics Covered

• PyTorch fundamentals
• Tensors and autograd
• Building models with nn.Module
• PyTorch data loading: Dataset, DataLoader
• Custom datasets creation
• Training loops in PyTorch
• GPU acceleration with PyTorch
• TorchVision for computer vision
• TorchText for NLP
• PyTorch Lightning for cleaner code
• Model checkpointing
• TensorBoard with PyTorch
• PyTorch vs TensorFlow comparison

Topics Covered

• Limitations of fully connected networks for images
• Convolution operation: filters and feature maps
• Padding and stride
• Pooling layers: max pooling, average pooling
• CNN architecture components
• LeNet architecture
• AlexNet and ImageNet revolution
• VGGNet: deep and simple
• Inception/GoogLeNet: multi-scale features
• ResNet: residual connections and skip connections
• Transfer learning with pretrained models
• Fine-tuning strategies
• Data augmentation for images

Topics Covered

• Object detection: R-CNN, Fast R-CNN, Faster R-CNN
• YOLO (You Only Look Once): real-time detection
• SSD (Single Shot Detector)
• Semantic segmentation: FCN, U-Net
• Instance segmentation: Mask R-CNN
• Face recognition and verification
• Siamese networks
• Image generation introduction
• Style transfer with CNNs
• OpenCV for computer vision
• Image preprocessing techniques
• Handling class imbalance in CV
• Model optimization for edge devices

Topics Covered

• NLP pipeline and challenges
• Text preprocessing: tokenization, lowercasing
• Stop words removal and stemming/lemmatization
• Bag of Words (BoW) and TF-IDF review
• Word embeddings: Word2Vec (CBOW, Skip-gram)
• GloVe embeddings
• FastText embeddings
• Using pretrained embeddings
• Text classification with embeddings
• Named Entity Recognition (NER)
• Part-of-Speech (POS) tagging
• Dependency parsing basics
• spaCy for NLP
• Why attention replaced recurrence: the honest one-lesson history

Topics Covered

• Transformer architecture: self-attention
• Multi-head attention
• Positional encoding
• BERT: Bidirectional Encoder Representations
• GPT: Generative Pre-trained Transformer
• T5, RoBERTa, ALBERT
• Hugging Face Transformers library
• Fine-tuning BERT for classification
• Zero-shot and few-shot learning
• Prompt engineering basics
• Modern NLP pipeline
• Production NLP systems

Topics Covered

• How large language models actually work: tokens, attention, training and alignment in honest depth
• The frontier landscape read soberly: GPT-class, Claude-class and open-weight Llama-class models
• Prompting as engineering: structure, iteration and evaluation harnesses
• Retrieval-augmented generation: embeddings, vector search, and when plain search wins
• Fine-tuning and adapters: what they change, what they cost, when they are wrong
• Structured output and function calling: making models act reliably
• Generative image models: how diffusion works and where it fits
• Evaluating generated output without fooling yourself
• Cost, latency and caching: the engineering triangle of GenAI

Topics Covered

• ML model lifecycle
• Model serialization: pickle, joblib
• Flask for ML APIs
• FastAPI for high-performance APIs
• REST API design for ML
• Input validation and preprocessing
• Model serving basics
• Docker for ML applications
• Creating ML microservices
• Model versioning strategies
• A/B testing deployed models
• Monitoring model performance
• Experiment tracking: knowing which run produced which model
• Serving a model behind an API, measured under load

Topics Covered

• What is MLOps? DevOps for ML
• ML system architecture
• ML pipeline orchestration
• Kubeflow for ML workflows
• Apache Airflow for ML pipelines
• Feature stores: Feast, Tecton
• Model registry: MLflow, DVC
• Experiment tracking at scale
• Metadata management
• Data lineage and provenance
• Continuous training (CT)
• Automated retraining pipelines
• Model governance
• Monitoring and drift: knowing when your model quietly went stale

Topics Covered

• Ethics in AI/ML
• Bias in machine learning
• Fairness metrics and definitions
• Detecting and mitigating bias
• Interpretable vs explainable AI
• LIME and SHAP for explanations
• Privacy in ML: differential privacy
• Federated learning for privacy
• Adversarial attacks on ML models
• Adversarial training for robustness

• Regulatory compliance: GDPR, AI Act
• Responsible AI frameworks
• Ethical considerations in deployment
• Explainability tools in practice: SHAP-style reasoning about model decisions
• Capstone planning: choosing a project worth defending

Topics Covered

• Two focused weeks: data, model, evaluation, deployment
• Instructor checkpoints and honest scope cuts
• Error analysis before excitement: the metric chosen and defended
• The capstone served behind an API with tracked experiments
• A model card a skeptic could audit

Topics Covered

• ML interview preparation strategy
• Coding interviews for ML roles
• ML theory and concepts questions
• System design for ML systems
• Take-home assignments approach
• Portfolio projects presentation
• GitHub for ML engineers
• Networking in AI community
• Conferences and meetups
• Building personal brand
• Salary negotiation for ML roles
• Career paths in AI/ML
• Walking through your own capstone the way interviews actually probe it

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: reinforcement learning, ML infrastructure, research
• Where our agents and Codex/Claude Code courses continue the modern layer