🤖 WDF Universe · Artificial Intelligence · Course ID AI-001

AI Master Class

Ten modules in four phases: from the mathematics behind machine learning to transformers, large language models, autonomous agents and responsible AI. Learn how AI really works, build it, and govern it well.

🆓 Free Course🎥 Video + Audio + Text📜 Certificate on Completion 🌐 Multilingual📱 Mobile Accessible🗓️ 24-Week Guided Path

🗺️ Explore the Roadmap Enrol Free Now → See All 10 Modules

Phases

Modules

Weeks, Self-Paced

Hands-On Labs

Learning Outcomes

Free

For Every Learner

Why It Matters

Why Learn AI Properly, Not Just Use It

🤖 AISO Summary · Why the AI Master Class matters. AI is moving from laboratories into healthcare, finance, farming and public services. Learners who understand how it works, not only how to use it, can build it, question it and govern it. The AI Master Class takes you from the mathematics of machine learning to autonomous agents and responsible deployment, in four phases and ten modules.

Most people meet AI as a finished product: a chatbot, a recommendation, a photo filter. Like an audience at a magic show, they accept the result without asking how the trick is done. That is harmless until AI starts to diagnose illness, approve loans or run public infrastructure. Then not understanding the mechanics becomes a risk. This course takes you backstage: it explains the mathematics, the models and the machinery, and then asks you to build and govern them yourself.

The skills landscape is moving quickly. In the United States, 2.5% of all job postings now mention AI skills, up 55% in a year (Lightcast, summarising the Stanford AI Index 2026), and the World Economic Forum’s Future of Jobs Report 2025 expects 39% of workers’ core skills to change by 2030.

Education has not kept pace. A large study of 3,395 computer-science syllabi from India’s top-100 engineering institutions found substantive AI-ethics content in only 75 of them (2.21%), and in 60% of those it was covered in a single lecture. The AI Master Class builds responsible practice into the technical sequence instead of leaving it as an afterthought.

This is also part of WDF Universe’s wider mission. The same technologies taught here (IoT sensors, drones, AI and cloud computing) power Knowledge Agriculture®, and learning them supports SDG-4 Quality Education, SDG-8 Decent Work and SDG-2 Zero Hunger.

2.5%

of US job postings mention AI skills, up 55% in a year.

Lightcast on the Stanford AI Index 2026

39%

of workers’ core skills are expected to change by 2030.

World Economic Forum, Future of Jobs 2025

2.21%

of 3,395 CS syllabi in India’s top-100 engineering institutions had substantive AI ethics.

Mittal et al. study, covered in the course syllabus

24 wks

is the suggested path from first principles to a deployed agent.

AI Master Class learning roadmap

Who is this course for?

🎓 Students and graduates in computing and engineering💻 Developers, analysts and data professionals🚀 Entrepreneurs and technology leaders🏫 Educators and curious learners with basic Python

What You Will Achieve

Eight Outcomes You Can Show

🤖 AISO Summary · Learning outcomes. By the end of the AI Master Class you will be able to explain how AI works, build and evaluate machine-learning and deep-learning models, adapt large language models, engineer secure prompts, create a working AI agent and audit AI systems for bias, transparency and environmental cost.

Explain how AI really works

Describe the mathematics and computing behind modern AI, from gradient descent to attention, without relying on jargon.

Build and evaluate models

Train classical machine-learning models and judge them honestly with precision, recall and F1.

Construct neural networks

Build and train networks in PyTorch or TensorFlow and understand backpropagation.

Process language and images

Turn text into vectors and images into matrices, then classify, detect and translate.

Adapt large language models

Understand transformers, fine-tune with LoRA and ground answers with retrieval (RAG).

Engineer prompts securely

Write structured prompts, use chain-of-thought and defend against prompt injection.

Create an AI agent

Design, build and deploy an agent that reasons, uses tools and remembers context.

Audit and govern AI

Test for bias, explain predictions, monitor drift and apply recognised governance frameworks.

Your Course Coordinators

Meet the Coordinators

🤖 AISO Summary · Course coordinators. The AI Master Class (Course ID AI-001) is coordinated by Prof. Dr. Sir Hari Om Srivastava, an authority in sustainable agriculture and knowledge systems, and Aprajita Srivastava, an AWS-certified Senior Cloud Engineer and published researcher in artificial intelligence, cloud computing, data platforms and ethical AI.

Prof. Dr. Sir Hari Om Srivastava

Course Coordinator

Prof. Dr. Sir Hari Om Srivastava (KStJ, D.Litt, Ph.D) is a scientist, inventor, and visionary leader whose work across six decades has shaped broadcasting technology, transformed agricultural knowledge systems, and extended the reach of education to millions. He is widely recognised as the "Father of Community Radio in India" and is the primary inventor of 'Knowledge Agriculture®' — a framework trademarked by the Government of India under Scientific and Technological Innovation. Ranked among the "Top 10 Indian Scientists" by Indian Amino in 2021 and declared "Scientist of the Month" by the Science Center, Gujarat in May 2020, he stands as one of India's most decorated and prolific scientific minds. He led large-scale initiatives in India and Ethiopia that delivered practical agricultural knowledge to hundreds of farmers.

Full profile ↗

Aprajita Srivastava

Course Coordinator

An AWS-certified Senior Cloud Engineer and a published researcher in artificial intelligence, cloud computing, data platforms and ethical AI. Her work applies modern analytics and cloud-native technologies to sustainable agriculture and social impact. She is also the Honorary Director of the World Development Foundation (WDF), where she directs research, educational initiatives and development projects aligned with the United Nations Sustainable Development Goals.

🏆 She received the Outstanding Scientist Award (AI/Education), 2026, from Scientific Laurels.

Full profile ↗

Course Structure

Four Phases, Ten Modules, One Path

🤖 AISO Summary · The AI Master Class roadmap. The journey moves from static rules to autonomous logic: Phase 1 builds mathematical and statistical foundations (Modules 1–2); Phase 2 gives machines the ability to see and read (Modules 3–5); Phase 3 covers generation with transformers and LLMs (Modules 6–8); Phase 4 builds autonomous agents and aligns them with human values (Modules 9–10).

Phase 1Technical & Mathematical FoundationsFoundations · Weeks 1–4

1AI Foundations & Computational Mathematics
2Classical Machine Learning & Data Methodology

Phase 2Deep Learning & Core ArchitecturesPerception · Weeks 5–12

3Deep Learning & Neural Network Fundamentals
4Natural Language Processing & Sequence Models
5Computer Vision & Visual Processing

Phase 3Advanced AI & Generative SystemsGeneration · Weeks 13–18

6Transformer Architectures & Attention Mechanisms
7Generative AI & Large Language Models
8Prompt Engineering & Model Alignment

Phase 4Autonomous Agents & Production DeploymentAutonomy & Governance · Weeks 19–24

9Agentic AI Systems & Tool Integration
10Responsible AI, Governance & Production Deployment

The 24-week learning path

Weeks 1–4FoundationsMaths, Python, data structures, classical ML

Weeks 5–12PerceptionDeep learning, PyTorch/TensorFlow, NLP and vision

Weeks 13–18GenerationTransformers, LLM fine-tuning, RAG, prompting

Weeks 19–24Autonomy & GovernanceAgents, RL, MLOps, ethical deployment

Key advice: build to learn. Establish a baseline, measure the outcome, and ship a system whose limitations you understand mathematically.

Course Catalogue

The 10 Modules in Detail

🤖 AISO Summary · The 10 modules. Each module lists what you will learn, a hands-on lab, the outcome you will be able to demonstrate, why it matters, curated videos and authoritative reading. Video and reading links point to courses, papers and documentation from universities, standards bodies and the original authors.

Module 1 of 10

AI Foundations & Computational Mathematics

Start with what AI is, where it came from and why it is worth mastering, then build the mathematical and computational base that every later module relies on.

📘 What you will learn

What AI is and how it works: systems that map inputs to goals through optimisation, statistical inference, search and pattern recognition
The AI family tree: artificial intelligence → machine learning → deep learning → neural networks → generative AI → large language models → agentic AI
A short history: the 1956 Dartmouth workshop, the AI winters, the GPU-driven deep-learning revolution after 2012, the 2017 transformer and the generative-AI boom
Symbolic and sub-symbolic AI, reasoning as search, and why the combinatorial explosion made brute-force logic impractical
Mathematics in plain language: vectors and tensors, linear algebra, multivariable calculus and gradients, probability and statistics
Python for AI and classical search: depth-first search, A* and minimax
AI buzzwords decoded: ANI, AGI and ASI; tokens and context windows; hallucination; fine-tuning; quantisation

🔬 Hands-on lab

Compare a rule-based (symbolic) lookup with a vector-similarity search on the same query, and explain when each one wins.

✅ You will be able to

Explain what AI is, trace its history, and describe, without jargon, how data, mathematics and computing power fit together.

💡 Why it matters

Treating AI as magic is risky. When you understand the mathematics, a model that fails stops being a broken magic box and becomes a problem you can diagnose.

🎥 Watch

Essence of Linear Algebra (full series)3Blue1Brown · PlaylistOpen playlist on YouTube ↗

📚 Authoritative reading & tools

Module 2 of 10

Classical Machine Learning & Data Methodology

Learn how machines find patterns in data, and how to judge honestly whether a model works. These methods are still the right tool for many real problems.

📘 What you will learn

The learning paradigms: supervised (classification and regression), unsupervised (clustering, PCA), reinforcement and transfer learning
Statistical classifiers and their trade-offs: decision trees, k-nearest neighbours, naive Bayes and support vector machines
Optimisation and search: gradient descent, heuristics and adversarial search
Reasoning under uncertainty: Bayesian networks, expectation-maximisation, Markov decision processes, hidden Markov models and Kalman filters
Evaluating models: MSE and RMSE, accuracy, precision, recall, F1 score and the confusion matrix
Data methodology: clean data, train/test splits, overfitting and why a good score on the wrong data proves nothing

🔬 Hands-on lab

Train classifiers with the no-code Orange Data Mining tool, then reproduce them in Python with scikit-learn and compare accuracy, speed and memory as the data grows.

✅ You will be able to

Choose a suitable classical model, train it, evaluate it with the right metric and explain its strengths and limits.

💡 Why it matters

Every deep-learning project starts from a simple baseline. Classical methods are fast, explainable and often good enough.

🎥 Watch

More: Full CS229 lecture playlist ↗

📚 Authoritative reading & tools

Module 3 of 10

Deep Learning & Neural Network Fundamentals

Build neural networks from the ground up, from a single neuron to multi-layer networks, and see exactly how a network learns from its mistakes.

📘 What you will learn

Artificial neurons, layers, weights and biases; activation functions (sigmoid, ReLU, softmax)
From perceptrons to multi-layer networks: why a single layer cannot solve XOR, and what the universal approximation theorem says
Loss functions, gradient descent and backpropagation (the chain rule at work)
Training problems: vanishing and exploding gradients, and the “dying ReLU”
Tensors and computational graphs in PyTorch and TensorFlow/Keras

🔬 Hands-on lab

Build and train a neural network in PyTorch or Keras, experiment in TensorFlow Playground and plot gradient sizes layer by layer across training.

✅ You will be able to

Build, train and debug a small neural network, and explain in plain words what backpropagation does.

💡 Why it matters

The real catalyst for deep learning was hardware. GPUs made deep, many-layered networks practical, and every later module builds on this one.

🎥 Watch

More: Gradient descent, how neural networks learn (chapter 2) ↗ · What is backpropagation really doing? (chapter 3) ↗

📚 Authoritative reading & tools

Neural Networks and Deep Learning (Michael Nielsen)Free book
PyTorch tutorialsDocs
TensorFlow PlaygroundInteractive

Module 4 of 10

Natural Language Processing & Sequence Models

See how machines turn human language into numbers they can compute with, and why that shift made modern language AI possible.

📘 What you will learn

Text processing: tokenisation and byte-pair encoding (BPE)
Word embeddings (Word2Vec, GloVe): meaning as geometry, where similar words sit close together and vector arithmetic captures relationships
Core applications: text classification, named-entity recognition, sentiment analysis and machine translation
Recurrent networks and LSTMs, and why long-range memory was their weak point
Language modelling as next-token prediction, and the path from rules and n-grams to embeddings, RNNs and transformers
Word-sense disambiguation: why the same word can mean different things in different sentences

🔬 Hands-on lab

Fine-tune a small pre-trained language model (for example BERT) on a text dataset of your own using Hugging Face Transformers, and inspect where the model pays attention.

✅ You will be able to

Represent text as vectors, build and evaluate a basic NLP pipeline, and explain why embeddings changed the field.

💡 Why it matters

Machines only understand numbers. Mapping language into a mathematical space is what lets software search, translate and summarise text.

🎥 Watch

📚 Authoritative reading & tools

Module 5 of 10

Computer Vision & Visual Processing

Teach machines to see. Start with how an image is stored as numbers, then move from classical image processing to convolutional neural networks.

📘 What you will learn

Digital image fundamentals: pixels, grayscale and RGB channels, resolution and matrix representation
Classical processing: filtering, edge detection, thresholding and morphological operations with OpenCV
Why learn classical methods first: they are explainable, and they cannot learn shortcuts that a deep network might
Convolutional neural networks: kernels, pooling and the hierarchy from edges to shapes to whole objects
Recognition tasks: classification, object detection and semantic segmentation, measured with intersection over union (IoU)
Machine perception: fusing camera, lidar, radar and sonar data, and the limits of affective computing

🔬 Hands-on lab

Load, resize, convert and threshold images with OpenCV, then train a CNN classifier. An optional advanced lab combines visual tracking with audio sentiment to drive a simulated avatar.

✅ You will be able to

Build a basic vision pipeline and judge when a classical method is better than a deep network.

💡 Why it matters

A medical-imaging model once learned to detect the ruler placed beside tumours rather than the tumours themselves. Knowing the pixels protects you from shortcuts like this.

🎥 Watch

**Stanford CS231n, Lecture 1: Introduction to Convolutional Neural Networks for Visual Recognition**Stanford University School of EngineeringWatch on YouTube ↗

More: Full CS231n lecture playlist (Spring 2017) ↗

📚 Authoritative reading & tools

CS231n course notesCourse notes
OpenCV documentationDocs

Module 6 of 10

Transformer Architectures & Attention Mechanisms

Open the engine block of modern AI. The transformer replaced word-by-word reading with attention that looks at a whole passage at once.

📘 What you will learn

Why sequential models hit a wall, and what the 2017 transformer changed
Self-attention: every word weighs its relevance to every other word, however far apart they are
Multi-head attention and encoder/decoder blocks
Self-supervised pre-training by next-token prediction on very large text collections
Scaling and efficiency: computational-graph optimisation and parallel training

🔬 Hands-on lab

Implement a small GPT-style transformer block by block and visualise its attention patterns.

✅ You will be able to

Explain self-attention step by step and implement a minimal transformer.

💡 Why it matters

Chatbots, code assistants and translation systems are built on this one idea, so understanding it is the key to understanding them.

🎥 Watch

More: Let’s build GPT: from scratch, in code, spelled out (Karpathy) ↗

📚 Authoritative reading & tools

Module 7 of 10

Generative AI & Large Language Models

Learn how large language models are built, trained and adapted, and how to make them more reliable by connecting them to real documents.

📘 What you will learn

Discriminative and generative models: one learns the boundary between classes, the other learns the distribution so it can create new examples
Foundation models and the training pipeline: pre-training, instruction tuning and alignment (RLHF, DPO)
Adapting models: supervised fine-tuning, LoRA and QLoRA, and quantisation for faster, cheaper deployment
Retrieval-augmented generation (RAG): an “open-book exam” for the model, using embeddings and vector databases such as FAISS and Pinecone
Multimodal generation of text, images, audio and video, and the tools used today
Hallucination: why a model that predicts the next likely word can state falsehoods with confidence

🔬 Hands-on lab

Fine-tune an open-weights model with LoRA and measure hallucination before and after. Then build a document question-answering chain with RAG using LangChain.

✅ You will be able to

Explain how an LLM is trained, adapt one to a task and ground its answers in retrieved documents.

💡 Why it matters

A standard LLM is like a student sitting a closed-book exam: it answers from memory. RAG lets it consult verified sources first.

🎥 Watch

More: Large Language Models explained briefly (3Blue1Brown) ↗

📚 Authoritative reading & tools

Module 8 of 10

Prompt Engineering & Model Alignment

Learn to steer language models with clear instructions, and to protect applications from people who try to make them misbehave.

📘 What you will learn

Prompting strategies: zero-shot, few-shot, system instructions and chain-of-thought reasoning
Prompt structure: role, context, constraints and structured output such as JSON or Markdown
Reasoning with search: how combining step-by-step generation with Monte Carlo Tree Search (rStar-Math) helps small models rival much larger ones on maths benchmarks
Alignment: how reinforcement learning from human feedback (RLHF) shapes a helpful, honest and safer assistant
Adversarial risks: prompt injection and jailbreaking
Defences: guardrails, output verification and formal verification, for example translating a proof into the Lean language so that a compiler can check it

🔬 Hands-on lab

Build a small automated red-team suite in Python that tests a language model against prompt-injection payloads and records where its safety boundaries fail.

✅ You will be able to

Write reliable, structured prompts and defend an LLM application against common injection attacks.

💡 Why it matters

Natural language is now a way to program machines. That brings new security risks: text hidden in a web page or document can try to take control of a model.

🎥 Watch

📚 Authoritative reading & tools

Module 9 of 10

Agentic AI Systems & Tool Integration

Move beyond chatbots to agents that observe, plan, act and learn. This module ends with your capstone project: a working AI agent.

📘 What you will learn

What an agent is: software that perceives its environment, remembers, plans and acts through tools to reach a goal
The agentic loop: perception, memory (short and long term), planning and action
ReAct: interleaving reasoning with actions; tool integration with web search, code sandboxes and external APIs
Memory and retrieval with vector databases (Pinecone, FAISS)
Planning and search: Monte Carlo Tree Search, best-of-N sampling and self-reflection
Decision theory: rational agents, expected utility, MDPs and POMDPs, and multi-agent game theory
Reinforcement learning: Q-learning, PPO and Soft Actor-Critic. Reward a game agent for eating a pellet and penalise it for hitting a ghost, and it discovers its own strategy
Real systems and limits: from Deep Blue (1997) to AlphaGo and DeepMind’s SIMA, and the limits of context windows, latency and scope

🔬 Hands-on lab

Capstone: design, build and deploy a conversational AI agent with the Google Gemini API or LangChain, using at least one external tool and a memory store.

✅ You will be able to

Build an agent that reasons step by step, calls tools and keeps context across a conversation.

💡 Why it matters

When AI can act, not just answer, mistakes have real consequences, which is why this module leads straight into responsible deployment.

🎥 Watch

📚 Authoritative reading & tools

Module 10 of 10

Responsible AI, Governance & Production Deployment

Treat ethics as an engineering requirement. Learn to audit AI for bias, explain its decisions, deploy it responsibly and keep watching it after launch.

📘 What you will learn

Algorithmic bias and fairness: distributive, representational and procedural fairness; why not every fairness definition can be satisfied at once
Privacy and intellectual property: differential privacy, data statements and copyright questions
Transparency: black-box models, and explainability tools such as SHAP, LIME and dictionary learning
Misinformation, misuse and the human cost of automation, including the risk of “de-skilling”
Energy and environment: electricity, cooling water and hardware demands of AI infrastructure
Governance frameworks: the EU AI Act, NIST AI Risk Management Framework, OECD AI Principles, UNESCO Recommendation on the Ethics of AI and India’s NITI Aayog responsible-AI strategy
Production deployment: APIs, model registries, monitoring, data drift and an “assume it will fail” design
Global South context: bias linked to caste and religion, biometric systems such as Aadhaar, and AI in agriculture and food distribution

🔬 Hands-on lab

Audit a biased credit dataset with Fairlearn, compare demographic parity and equalised odds, explain predictions with SHAP or LIME, and estimate the energy and water footprint of a training run.

✅ You will be able to

Audit a model for bias, explain its decisions, plan monitoring for data drift and map a project to recognised governance frameworks.

💡 Why it matters

A study of 3,395 computer-science syllabi in India’s top-100 engineering institutions found substantive AI-ethics content in only 75 of them (2.21%). Closing that gap is part of this course’s purpose.

📚 Authoritative reading & tools

NIST AI Risk Management FrameworkFramework
EU Artificial Intelligence Act, Regulation (EU) 2024/1689Law
OECD AI PrinciplesPrinciples
UNESCO Recommendation on the Ethics of Artificial IntelligenceStandard
Fairlearn: fairness assessment and mitigationToolkit
SHAP documentationToolkit
ACM/IEEE Computer Science Curricula 2023Curriculum
IEA: Energy and AIReport

Companion Materials

Listen, Watch and Download

🤖 AISO Summary · Companion materials. An audio overview (“How AI Really Works: Under the Hood”), a 15-slide deck (“Architecting Intelligence”) and the full syllabus give you three more ways into the course: listen on the move, skim visually, or study the detailed curriculum.

🎧 Audio overview: How AI Really Works, Under the Hood

Audio · about 21 minutes

Six ideas from the conversation

Don’t just watch the magic showTreating AI as magic is the riskiest thing you can do when it is diagnosing patients, moving money and running infrastructure. Learn the stage mechanics, and the trapdoors become visible.
From deduction to inductionClassical AI was given the rules of Sudoku. Machine learning is given the starting boards and the solved answers and works out the rules itself.
Intuition for the mathematicsA tensor is just a way of organising numbers in more dimensions. Gradient descent is like walking downhill blindfolded, feeling the slope, until the ground is flat: the lowest point of the loss.
Why classical vision still mattersA deep network can take a shortcut, such as spotting the ruler in the photo instead of the tumour. Classical image processing is explainable and forces you to understand the pixels.
Language as geometryWords become coordinates, so “king − man + woman” lands near “queen”. Attention lets a model weigh every word against every other, and RAG gives it an open book to consult.
Agents, ethics and the audience questionRewards teach a game agent to avoid ghosts. Ethics must be an engineering constraint: document your data, watch for data drift, and keep asking whether you are upgrading your own judgement or outsourcing it.

🖼️ Slide deck: Architecting Intelligence

15 slides · roadmap, modules and the modern AI stack

Slide 1: Architecting Intelligence, the 10-module AI Masterclass

Slide 1 of 15

Module numbers on slides 6–10 follow the original slide deck. In this course, Natural Language Processing is Module 4 and Computer Vision is Module 5; generative AI, LLMs and RAG are spread across Modules 6–8.

⬇ Download slide deck (PPTX, 18 MB) ⬇ Download full syllabus (DOCX)

Practise & Prove It

Projects That Build Real Skill

🤖 AISO Summary · Practice and assessment. Instead of relying only on multiple-choice quizzes, the course uses applied projects: a bias audit, an ethical impact assessment, a red-team lab, an explainability brief and an agent capstone. Complete the course evaluation to receive your WDF Universe certificate.

Algorithmic bias audit

Audit a real-world style dataset such as credit scoring. Measure fairness across groups, document sample-size gaps, apply a mitigation and explain the trade-offs between fairness definitions.

Ethical impact assessment

Before the capstone, assess privacy risks, the power and cooling-water footprint, job impacts and local context, and describe fallback safety protocols for failures.

Red-team and safety lab

Attack a language model with prompt injections and jailbreak attempts, record where its boundaries fail and engineer defences.

Explainability & policy brief

Apply SHAP or LIME to a black-box model, look for shortcut learning, and turn your findings into a brief a compliance auditor could use.

★

Capstone: your own AI agent

Design, build and deploy a conversational agent with the Google Gemini API or LangChain, with tool use, memory and documented limitations.

Quick Reference

Frequently Asked Questions

Is the AI Master Class free?▼

Yes. Like all courses on WDF Universe, the AI Master Class is free to learners worldwide. Learners who complete the course evaluation receive a WDF Universe certificate.

What do I need to know before starting?▼

The course is designed for learners from beginner to advanced level. You need basic Python programming and high-school mathematics. Module 1 and Module 2 rebuild the mathematics and statistics you need in plain language.

How long does it take?▼

The suggested learning path is 24 weeks: Foundations in weeks 1–4, Perception in weeks 5–12, Generation in weeks 13–18 and Autonomy & Governance in weeks 19–24. The course is self-paced, so you can go faster or slower.

Do I need to be good at advanced mathematics?▼

You do not need to derive equations from memory. You do need an intuitive understanding of what the mathematics is doing, for example how a gradient tells a model which way to adjust. The course builds that intuition first and the formulas second.

Which tools will I use?▼

Python, scikit-learn and the no-code Orange Data Mining tool for classical machine learning; PyTorch and TensorFlow/Keras for deep learning; OpenCV for vision; Hugging Face Transformers for language models; LangChain and the Google Gemini API for retrieval and agents; and Fairlearn and SHAP for fairness and explainability.

Why does a course about AI include ethics and governance?▼

Because responsible deployment is part of engineering. A study of 3,395 computer-science syllabi in India’s top-100 engineering institutions found substantive AI-ethics content in only 2.21% of them. Module 10, and the bias audits, red-teaming and impact assessments throughout the course, are designed to close that gap.

In which languages can I learn?▼

Use the language selector at the top of the page to translate the course text into your preferred language through built-in translation. Lessons are available in video, audio and text, and work on mobile devices.

Start Learning. Build with AI.

Join learners worldwide in a free, certified course that takes you from first principles to a working, responsibly governed AI agent.

🎓 Enrol Free Now → Browse All Courses Download Syllabus