IA Certificate Course

Lesson 1 • History and Evolution of AI

Traces AI development from symbolic logic to modern neural networks. Provides historical context that explains why current techniques dominate the field.

Lesson 2 • Core AI Paradigms

Introduces supervised, unsupervised, and reinforcement learning as the three foundational paradigms. Connects each paradigm to the problem types it solves.

Lesson 3 • AI in Industry and Society

Maps AI applications across healthcare, finance, logistics, and media. Grounds abstract concepts in tangible use cases students will encounter professionally.

Lesson 1 • Exploratory Data Analysis

Applies statistical summaries and visualisations to uncover patterns before modelling. Builds the habit of understanding data distributions prior to algorithm selection.

Lesson 2 • Linear Algebra for AI

Covers vectors, matrices, and tensor operations central to neural network computation. Connects matrix multiplication directly to layer transformations in models.

Lesson 3 • Calculus and Optimisation Basics

Introduces derivatives, gradients, and the chain rule as tools for training models. Links these concepts to gradient descent optimisation used in every AI model.

Lesson 4 • Probability and Statistics Fundamentals

Teaches probability distributions, Bayes' theorem, and statistical inference. These tools underpin model uncertainty estimation and evaluation metrics.

Lesson 1 • Classification Algorithms

Introduces logistic regression, decision trees, and support vector machines for categorical prediction. Compares decision boundaries and computational trade-offs.

Lesson 2 • Clustering and Dimensionality Reduction

Covers k-means, hierarchical clustering, PCA, and t-SNE for unsupervised pattern discovery. Links dimensionality reduction to visualisation and feature engineering.

Lesson 3 • Regression Techniques

Covers linear and polynomial regression for continuous output prediction. Establishes the loss-minimisation framework reused in all subsequent algorithms.

Lesson 4 • Model Evaluation and Selection

Teaches cross-validation, bias-variance trade-off, and performance metrics for fair model comparison. Ensures students can justify algorithm choices with quantitative evidence.

Lesson 1 • Feedforward Networks and Backpropagation

Explains multi-layer perceptron architecture and the backpropagation algorithm. Connects gradient flow to the calculus concepts introduced in Chapter 2.

Lesson 2 • Perceptrons and Activation Functions

Introduces the biological neuron analogy, the perceptron model, and activation functions. Establishes the computational unit that all deeper architectures build upon.

Lesson 3 • Recurrent Neural Networks and Sequences

Covers RNNs, LSTMs, and GRUs for sequential and time-series data modelling. Explains how gating mechanisms solve the vanishing gradient problem in sequences.

Lesson 4 • Convolutional Neural Networks

Introduces convolution, pooling, and CNN architectures for image and spatial data. Demonstrates how local feature extraction enables translation invariance.

Lesson 1 • Attention Mechanisms and Transformers

Introduces self-attention, multi-head attention, and the transformer architecture. Explains why transformers replaced RNNs as the dominant NLP backbone.

Lesson 2 • NLP Applications and Pipelines

Applies NLP models to sentiment analysis, named entity recognition, and summarisation. Integrates preprocessing, modelling, and postprocessing into end-to-end pipelines.

Lesson 3 • Pre-trained Language Models

Covers BERT, GPT, and T5 as foundational pre-trained models for NLP tasks. Demonstrates fine-tuning strategies that adapt general models to specific domains.

Lesson 4 • Text Preprocessing and Representation

Covers tokenisation, stemming, lemmatisation, and vector representations of text. Establishes the data pipeline that feeds all downstream NLP models.

Lesson 1 • Generative Adversarial Networks

Explains the generator-discriminator game and GAN training dynamics. Covers mode collapse, training instability, and practical mitigation strategies.

Lesson 2 • Diffusion Models and Multimodal AI

Introduces denoising diffusion models and vision-language models like CLIP. Positions these as the current frontier of generative and multimodal AI.

Lesson 3 • Variational Autoencoders

Covers the VAE latent space, reparameterisation trick, and evidence lower bound. Contrasts VAEs with GANs in terms of output diversity and training stability.

Lesson 4 • Semantic Segmentation and Pose Estimation

Introduces pixel-level labelling and human pose estimation as advanced vision tasks. Demonstrates how dense prediction differs from bounding-box detection.

Lesson 1 • Scalability and Infrastructure

Introduces cloud-based GPU clusters, distributed training, and auto-scaling for large models. Prepares students to make cost-performance trade-offs in production environments.

Lesson 2 • Model Packaging and Serving

Explains containerisation, REST APIs, and model registries for serving predictions. Connects packaging standards to consistent behaviour across environments.

Lesson 3 • Monitoring and Observability

Covers data drift, concept drift, and performance degradation detection in production. Teaches logging and alerting strategies that maintain model reliability over time.

Lesson 4 • ML Pipeline Design and Automation

Covers end-to-end pipeline components from data ingestion to model serving. Establishes reproducibility and automation as core production engineering values.

Lesson 1 • AI Governance and Regulatory Landscape

Surveys global AI governance frameworks, risk classification, and organisational accountability structures. Equips students to navigate compliance requirements without referencing specific legal codes.

Lesson 2 • Explainability and Interpretability

Covers SHAP, LIME, and attention visualisation as tools for model transparency. Explains why interpretability is essential for trust, debugging, and regulatory compliance.

Lesson 3 • AI Strategy and Business Alignment

Teaches how to build an AI roadmap, measure ROI, and align AI projects with business objectives. Closes the course by connecting technical skills to strategic decision-making.

Lesson 4 • Bias Sources and Fairness Metrics

Identifies data, algorithmic, and societal bias sources and quantifies them with fairness metrics. Connects bias detection to the preprocessing and evaluation skills from earlier chapters.