Skip to contentCourse
Professional Applied AI Program
From Python basics to agents running in production.
The full hands-on path from Python to production agents: eleven courses of projects, built around your job and your week.
567 topics107 h hands-on161 h in depthLifetime access
What you'll be able to do
- Build and ship a RAG assistant on your own data
- Design multi-agent systems with tools and guardrails
- Deploy and monitor an AI service with CI/CD
What's inside
Fundamentals (Python, NumPy, pandas, Maths)76 topics · 14 h
Python Foundations by Building30 topics · 6.1 h
B00 · Project Setup: your first AI repo2 topics · 0.5 h
- Set up Python, VS Code and a virtual environment15 min
- Notebook vs script, Git and GitHub15 min
B01 · Build: Expense Receipt Generator3 topics · 0.8 h
- Capture expenses from the user15 min
- Compute subtotal, tax, discount and total15 min
- Print a formatted receipt15 min
B02 · Build: Review Text Cleaner & Word Counter4 topics · 1 h
- Clean a raw customer review15 min
- Store words in lists, tuples and sets15 min
- Count word frequencies with a dictionary15 min
- Rewrite loops as comprehensions15 min
B03 · Build: Loan Eligibility Checker3 topics · 0.8 h
- Decide eligibility with if / elif / else15 min
- Handle missing values: truthiness and None15 min
- Replace an if-ladder with a lookup table15 min
B04 · Build: Batch Transaction Processor with Retries (loops)5 topics · 0.9 h
- Loop over transactions with for and range15 min
- Track position and pair lists: enumerate and zip10 min
- Skip and stop: continue, break, pass10 min
- Retry a flaky API with a while loop10 min
- Nested loops and why a dict is faster10 min
B05 · Build: Grade Book Toolkit (functions & modules)4 topics · 0.7 h
- Turn scripts into functions10 min
- Flexible arguments and lambda sort keys10 min
- Type hints, scope and recursion (glimpse)10 min
- Split into modules and use the standard library10 min
B06 · Build: Persistent Contacts & Logs (files, JSON, errors)3 topics · 0.5 h
- Save and load data as JSON and CSV10 min
- Handle errors without crashing10 min
- Replace print with logging10 min
B07 · Build: Bank Account System (classes)3 topics · 0.5 h
- Model an account with a class10 min
- Extend with inheritance and dataclasses10 min
- Make objects behave like built-ins (glimpse)10 min
B08 · Build: Live Currency Converter (APIs, JSON, generators)3 topics · 0.5 h
- Call a public API and parse nested JSON10 min
- Stream and batch with generators10 min
- Run calls concurrently with async (glimpse)10 min
NumPy & Linear Algebra by Building17 topics · 2.8 h
B09 · Build: Class Marks Analyzer with NumPy5 topics · 0.8 h
- Create arrays and inspect shape and dtype10 min
- Select students and subjects: indexing, slicing, masks10 min
- Summarise with aggregations along an axis10 min
- Standardise marks with broadcasting (z-scores)10 min
- Reshape, stack and save results10 min
B10 · Build: Movie Recommender from Ratings (vectors & similarity)4 topics · 0.7 h
- Users and movies as vectors10 min
- Find similar users with dot product and cosine similarity10 min
- Distances: Euclidean vs Manhattan vs cosine10 min
- Recommend top-5 movies with matrix multiplication10 min
B11 · Build: Photo Filter Studio (images as matrices and tensors)4 topics · 0.7 h
- Load an image as a 3-D array10 min
- Apply filters with array operations10 min
- Rotate points with a matrix (linear transformations)10 min
- Compress the image with SVD (glimpse of low-rank)10 min
B12 · Build: Probability Simulation Lab (dice, coins, conversions)4 topics · 0.7 h
- Simulate random events with a seeded generator10 min
- Expectation, variance and the Law of Large Numbers10 min
- Central Limit Theorem by simulation10 min
- Bayes' theorem with a disease-test simulation10 min
pandas & EDA by Building15 topics · 2.5 h
B13 · Build: Retail Sales Explorer (pandas basics)4 topics · 0.7 h
- Load and inspect a real dataset10 min
- Select, filter and sort10 min
- Create new columns and transform values10 min
- Count and describe categories10 min
B14 · Build: Data Cleaning Clinic (messy customer data)3 topics · 0.5 h
- Find and treat missing values10 min
- Fix types, duplicates and inconsistent labels10 min
- Detect and cap outliers10 min
B15 · Build: Customer Insights Report (grouping, joining, reshaping)5 topics · 0.8 h
- Group and aggregate10 min
- Join tables and stack files10 min
- Reshape: pivot tables, crosstab, melt and binning10 min
- Time-based analysis10 min
- Correlation: relationship, not cause10 min
B16 · Build: Visual EDA Story (Matplotlib & Seaborn)3 topics · 0.5 h
- Matplotlib essentials10 min
- Seaborn statistical plots10 min
- Write the insight summary10 min
Maths of Learning by Building14 topics · 2.6 h
B17 · Build: Fit a Line with Gradient Descent (calculus in action)5 topics · 0.8 h
- Define the model and the loss10 min
- See the derivative as slope (no hand calculus)10 min
- Train with gradient descent10 min
- Plot the loss curve and compare with library results10 min
- Momentum and Adam (glimpse)10 min
B18 · Build: Spam Classifier with Probabilities (sigmoid, softmax, cross-entropy)4 topics · 0.7 h
- Turn scores into probabilities with sigmoid10 min
- Measure surprise: log loss (cross-entropy)10 min
- Multi-class with softmax and temperature10 min
- Evaluate with a confusion matrix10 min
B19 · Build: See Your Data in 2-D (PCA)2 topics · 0.3 h
- Centre the data and build the covariance matrix10 min
- Find the main directions and project to 2-D10 min
B20 · Build: A/B Test the Checkout Page (statistics for decisions)2 topics · 0.3 h
- Compare conversion rates with a bootstrap10 min
- Run a significance test and read the p-value10 min
CAPSTONE · Loan Default EDA + Model from Scratch1 topics · 0.4 h
- End-to-end notebook25 min
EDA72 topics · 13 h
Exploratory Data Analysis by Building (two projects in parallel)72 topics · 13 h
E00 · Build: EDA Project Workspace2 topics · 0.5 h
- Set up the repo, data folders and EDA notebook template15 min
- Write the shared helper toolkit (first 3 functions)15 min
E01 · Build: Problem Framing Brief (one page per project)5 topics · 1 h
- Business problem, decision and success criteria15 min
- Cost of errors (cost matrix) and the metric it implies15 min
- Unit of analysis, target and problem type10 min
- Hypotheses to test10 min
- Data dictionary and leakage watch-list10 min
E02 · Build: Load, Structure & Integrity Report5 topics · 0.8 h
- Load data and confirm parsing10 min
- Join tables at the right grain (Project B)10 min
- Shape, grain, dtypes and column names10 min
- Keys, duplicates and repeated entities10 min
- Range, format and consistency checks10 min
E03 · Build: Data Quality & Missing-Value Log4 topics · 0.7 h
- Quantify missingness10 min
- Missingness patterns and missing by target/segment10 min
- Placeholder and invalid values10 min
- Data quality issue log and impact10 min
E04 · Build: Target Analysis6 topics · 1 h
- Validate the target10 min
Machine Learning (Supervised + Unsupervised)91 topics · 17 h
Machine Learning Foundations16 topics · 4 h
M00 · Build: ML Project Workspace (continuing from the EDA)2 topics · 0.5 h
- Reuse the EDA outputs and set up the ML repo15 min
- Start an experiment leaderboard15 min
M01 · Build: What Machine Learning Is (rules vs learning)5 topics · 1.3 h
- Traditional programming vs machine learning15 min
- Types of machine learning15 min
- ML problem types15 min
- Generalisation, overfitting and underfitting (build it)15 min
- No Free Lunch and when ML is not the answer15 min
M02 · Build: Project Understanding, Data Collection & Data Synthesis5 topics · 1.3 h
- From business problem to ML project (recap of EDA framing)15 min
- Data sources, collection and labels15 min
- What data synthesis is and when to use it15 min
- Generate synthetic data (3 methods)15 min
- Check synthetic data quality: fidelity, utility, privacy15 min
M03 · Build: Model-Ready Data & Baselines4 topics · 1 h
- X, y and the three-way split15 min
- Leakage-safe preprocessing pipeline (from EDA E13)15 min
- When scaling is needed and when it is not15 min
- Baselines you must beat15 min
Supervised Learning by Building45 topics · 8.1 h
AI Tools & Vibe Coding27 topics · 5.5 h
Part 1 · AI Coding Agents: Claude Code, Copilot, Cursor, Codex8 topics · 1.5 h
1.1 Coding agents1 topics · 0.2 h
- What coding agents are and which to use10 min
1.2 Set up the tools1 topics · 0.2 h
- Install and connect Claude Code, Copilot, Cursor and Codex10 min
1.3 Settings that matter3 topics · 0.5 h
- Models, reasoning effort and context10 min
- Permission modes and safety10 min
- Project instructions: CLAUDE.md, AGENTS.md, rules files10 min
1.4 The safe workflow1 topics · 0.2 h
- Plan → small steps → review → test → commit10 min
1.5 Security and cost1 topics · 0.2 h
- Secrets, risky code and spending10 min
1.6 Mini project1 topics · 0.3 h
- Same feature, two tools — compare20 min
Part 2 · Build Websites with Lovable (no-code / low-code)7 topics · 1.4 h
2.1 Lovable essentials3 topics · 0.5 h
- AI app builders and when to use them10 min
- All the important Lovable settings10 min
- Prompting an app builder10 min
2.2 Project: course-booking website4 topics · 0.9 h
Deep Learning51 topics · 10 h
Module 1 · Introduction to Deep Learning14 topics · 2.7 h
1.1 What is deep learning2 topics · 0.3 h
- AI vs machine learning vs deep learning10 min
- When to use deep learning (and when not)10 min
1.2 Neural network basics (ANN)3 topics · 0.5 h
- Neuron (perceptron), weights and bias10 min
- Layers and the forward pass10 min
- Activation functions10 min
1.3 How a network learns3 topics · 0.7 h
- Loss function: measuring the mistake10 min
- Gradient descent and backpropagation20 min
- Epochs, batches and optimisers10 min
1.4 Making networks generalise2 topics · 0.3 h
- Train / validation / test and overfitting10 min
- Dropout, early stopping and batch normalisation10 min
1.5 Types of neural networks1 topics · 0.2 h
- ANN, CNN, RNN/LSTM and Transformers — what each is for10 min
1.6 Deep learning frameworks2 topics · 0.3 h
- Tensors, GPU and Google Colab10 min
- PyTorch vs TensorFlow/Keras10 min
1.7 Mini exercise1 topics · 0.3 h
- Your first neural network on Fashion-MNIST (Keras and PyTorch)20 min
NLP & Transformers46 topics · 8.5 h
Module 1 · Introduction to NLP9 topics · 2 h
1.1 What is NLP2 topics · 0.5 h
- NLP in everyday life15 min
- The main NLP tasks (and which project covers them)15 min
1.2 Text preprocessing2 topics · 0.5 h
- Tokens, cleaning, stopwords15 min
- Stemming and lemmatisation15 min
1.3 Turning text into numbers2 topics · 0.5 h
- Bag of Words and TF-IDF15 min
- Word embeddings: meaning as vectors15 min
1.4 From RNNs to Transformers1 topics · 0.2 h
- Why sequence models were not enough10 min
1.5 NLP tools1 topics · 0.2 h
- NLTK, spaCy, scikit-learn and Hugging Face10 min
1.6 Mini exercise1 topics · 0.2 h
- Explore and represent a text dataset10 min
Module 2 · Transformers Explained9 topics · 1.5 h
2.1 Attention3 topics · 0.5 h
- Why attention: look at every word at once10 min
- Self-attention: Query, Key, Value10 min
- Multi-head attention and positional encoding10 min
2.2 The Transformer2 topics · 0.3 h
LLMs & Generative AI54 topics · 11 h
Module 1 · Introduction to Generative AI and LLMs8 topics · 2 h
1.1 What is generative AI1 topics · 0.3 h
- Generative vs predictive AI15 min
1.2 How LLMs work3 topics · 0.8 h
- Next-token prediction15 min
- Pretraining, instruction tuning and human feedback15 min
- Tokens, context window and model size15 min
1.3 The model landscape1 topics · 0.3 h
- Closed vs open models and how to choose15 min
1.4 Limits and responsible use1 topics · 0.3 h
- Hallucination, cut-off, bias and privacy15 min
1.5 Ways to use an LLM1 topics · 0.3 h
- API and local models15 min
1.6 Mini exercise1 topics · 0.3 h
- Cloud model vs local model15 min
Module 2 · Prompt Engineering8 topics · 1.8 h
2.1 Talking to an LLM2 topics · 0.5 h
- System, user and assistant messages15 min
- Temperature and max tokens15 min
2.2 Writing good prompts4 topics · 1 h
- Prompt structure: role, task, context, rules, format15 min
- Zero-shot and few-shot prompting
Agentic AI43 topics · 9 h
Module 1 · Introduction to AI Agents8 topics · 1.3 h
1.1 What is an AI agent2 topics · 0.3 h
- Chatbot vs workflow vs agent10 min
- The agent loop and ReAct10 min
1.2 Types and parts of an agent2 topics · 0.3 h
- Types of agents10 min
- The 5 parts: model, tools, memory, planning, guardrails10 min
1.3 Design patterns1 topics · 0.2 h
- Workflow patterns and agent patterns10 min
1.4 Control and safety1 topics · 0.2 h
- Autonomy levels and human-in-the-loop10 min
1.5 Frameworks1 topics · 0.2 h
- The framework landscape and how to choose10 min
1.6 Mini exercise1 topics · 0.2 h
- A ReAct agent from scratch10 min
Module 2 · Project 1: Business Operations Agent — Finance and Inventory (single agent with LangGraph)10 topics · 2 h
2.1 Understand the problem1 topics · 0.2 h
- Define the product and how we will judge it10 min
2.2 Data and test tasks2 topics · 0.3 h
- Create synthetic business data10 min
- Write the test task set10 min
Voice Agents52 topics · 9 h
Module 1 · Voice Agent Foundations14 topics · 2.7 h
1.1 What voice agents are1 topics · 0.3 h
- Inbound vs outbound calling agents and use cases15 min
1.2 The voice pipeline3 topics · 0.8 h
- STT → LLM → TTS (cascade) vs speech-to-speech15 min
- Speech-to-text (STT) essentials15 min
- Text-to-speech (TTS) essentials15 min
1.3 Audio and phone basics2 topics · 0.3 h
- Audio basics: sample rate, codecs, phone vs web audio10 min
- Telephony basics: numbers, PSTN, SIP, transfer, DTMF, voicemail10 min
1.4 Real-time conversation3 topics · 0.5 h
- Streaming, VAD, turn detection and interruptions10 min
- Latency budget10 min
- WebRTC vs WebSocket vs phone10 min
1.5 The LLM part for voice2 topics · 0.3 h
- Writing prompts for speech10 min
- Tools, knowledge and memory in a call10 min
1.6 Choosing the stack1 topics · 0.2 h
- Providers, cost per minute and Indian languages10 min
1.7 Rules and safety1 topics · 0.2 h
- Consent, recording, do-not-disturb and personal data10 min
1.8 Two ways to build1 topics · 0.2 h
Deployment & LLMOps43 topics · 7.5 h
Module 1 · Deployment Foundations9 topics · 1.8 h
1.1 From notebook to product1 topics · 0.3 h
- What deployment means and the parts of an AI app15 min
1.2 Web basics2 topics · 0.5 h
- HTTP, methods, status codes, JSON and REST15 min
- Testing APIs with Postman and Swagger15 min
1.3 Config and code2 topics · 0.4 h
- Environment variables and secrets15 min
- Git and GitHub team workflow10 min
1.4 Server basics1 topics · 0.2 h
- Terminal and Linux essentials10 min
1.5 Where to deploy1 topics · 0.2 h
- Hosting options and how to choose10 min
1.6 After launch1 topics · 0.2 h
- MLOps vs LLMOps: the life cycle10 min
1.7 Mini exercise1 topics · 0.2 h
- Call, test and document an API10 min
Module 2 · Project 1: Serve AI Models as an API (FastAPI)8 topics · 1.3 h
2.1 Starter project1 topics · 0.2 h
- Prepare the two models10 min
2.2 FastAPI basics1 topics · 0.2 h
Automation with n8n12 topics · 3 h
Part 1 · n8n Essentials and Settings7 topics · 1.7 h
1.1 Automation basics1 topics · 0.2 h
- What automation tools do and which to choose10 min
1.2 Set up n8n1 topics · 0.3 h
- n8n Cloud or self-host with Docker15 min
1.3 Core concepts2 topics · 0.5 h
- Triggers, nodes, items and expressions15 min
- Credentials, HTTP requests and webhooks15 min
1.4 Logic1 topics · 0.3 h
- IF, Switch, Merge, loops and the Code node15 min
1.5 AI in n8n1 topics · 0.3 h
- AI Agent node, chains, classification and extraction15 min
1.6 Reliable workflows1 topics · 0.3 h
- Errors, executions, sub-workflows, versions and security15 min
Part 2 · Project: AI Lead-to-Customer Automation (end to end)5 topics · 1.3 h
2.1 Plan1 topics · 0.3 h
- Design the automation15 min
2.2 Build2 topics · 0.5 h
- Capture, qualify and save leads15 min
- Personalised email, alerts and daily report15 min
2.3 AI assistant1 topics · 0.3 h
Classification: class balance (Project A)
10 min
Regression: target distribution (Project B)10 minTarget by key segments10 minTarget leakage check10 minProtected attributes and fair-lending check10 minE05 · Build: Univariate Analysis — Numerical Features5 topics · 0.8 h
- Summary statistics table10 min
- Distribution shape: histograms, KDE, box plots, ECDF10 min
- Skewness and transformations10 min
- Outlier flags with IQR and z-scores10 min
- Normality: when it matters and when it does not10 min
E06 · Build: Univariate Analysis — Categorical Features4 topics · 0.7 h
- Counts, proportions and cardinality10 min
- Clean inconsistent labels10 min
- Consolidate categories (domain logic first, bad rate second)10 min
- Identifier-like and high-cardinality fields10 min
E07 · Build: Bivariate & Multivariate Relationships8 topics · 1.3 h
- Numerical vs numerical10 min
- Categorical vs numerical10 min
- Categorical vs categorical10 min
- Correlation: Pearson, Spearman, Kendall and the heatmap10 min
- Multicollinearity and redundant features (VIF)10 min
- Significance vs effect size10 min
- Mutual information for non-linear relationships10 min
- Interactions and subgroup patterns10 min
E08 · Build: Outliers, Anomalies & Robustness3 topics · 0.5 h
- Multivariate outliers10 min
- Error, rare event or anomaly? Decide and treat10 min
- Before/after comparison and sensitivity check10 min
E09 · Build: Time-Based EDA (Project B focus)4 topics · 0.7 h
- Time coverage, frequency and gaps10 min
- Trend, seasonality and calendar effects10 min
- Lags, rolling windows and autocorrelation10 min
- Temporal splits and drift10 min
E10 · Build: Feature Engineering Plan & Leakage-Safe Preparation7 topics · 1.2 h
- Domain-based features10 min
- Validate engineered features (keep only what helps)10 min
- Date and time features (cyclical encoding)10 min
- Rare-category consolidation and encoding needs10 min
- Encode column by column (binary, ordinal, one-hot)10 min
- Scaling needs and stability across samples10 min
- Leakage-safe preparation plan10 min
E11 · Build: EDA for Segmentation (Project A without the target)7 topics · 1.2 h
- Scope of this mini-segmentation (read first)10 min
- Pick features and prepare them for distances10 min
- Find the number of groups and a 2-D view10 min
- Profile and name the personas (our own classes)10 min
- Bring the target back: bad rate per persona10 min
- Rule-based risk tiers (a multi-class label of our own)10 min
- Check stability and usefulness10 min
E12 · Build: Feature Shortlist & EDA Report6 topics · 1 h
- Information Value and Weight of Evidence (credit classification)10 min
- Quick baseline and importance caveats10 min
- Consensus feature shortlist10 min
- Automated profiling as a cross-check10 min
- Findings, limitations and stakeholder story10 min
- Data readiness checklist10 min
E13 · Build: Training-Ready Dataset Hand-off5 topics · 0.8 h
- Split first, then learn every transformation on train only10 min
- Build the preprocessing Pipeline (ColumnTransformer)10 min
- Imbalance strategy preview (train only)10 min
- Cost-based threshold preview (on validation, never on test)10 min
- Final validation checklist and export10 min
CAPSTONE · Two EDA Reports + Reusable EDA Toolkit1 topics · 0.3 h
- Deliver both EDAs from one toolkit20 min
S01 · Build: How Training Works (fit, loss, optimisation)5 topics · 0.9 h
- Parameters vs hyperparameters15 min
- Loss functions and what they reward10 min
- Optimisation: gradient descent, epochs and batches in practice10 min
- Learning curves: more data or a better model?10 min
- The scikit-learn API and how other libraries differ10 min
S02 · Build: An Evaluation Harness (used for every model)5 topics · 0.8 h
- Classification metrics from the confusion matrix10 min
- Ranking metrics: ROC-AUC and PR-AUC10 min
- Thresholds, costs and calibration10 min
- Regression metrics10 min
- Cross-validation done right10 min
S03 · Build: Choosing the Right Algorithm (decision guide)1 topics · 0.2 h
- A practical algorithm decision guide10 min
S04 · Build: Linear Models (the baselines everyone must master)3 topics · 0.5 h
- Logistic Regression | Linear Regression10 min
- Polynomial and interaction features10 min
- Regularisation: Ridge, Lasso, Elastic Net (and L1/L2 for logistic)10 min
S05 · Build: Distance, Probability & Margin Models4 topics · 0.7 h
- K-Nearest Neighbours (classifier | regressor)10 min
- Naive Bayes (classification)10 min
- LDA and QDA (glimpse)10 min
- Support Vector Machines (SVC | SVR)10 min
S06 · Build: Decision Trees1 topics · 0.2 h
- Decision Tree (classifier | regressor)10 min
S07 · Build: Bagging Ensembles2 topics · 0.3 h
- Random Forest (classifier | regressor)10 min
- Extra Trees (classifier | regressor)10 min
S08 · Build: Boosting (the tabular state of the art)5 topics · 0.8 h
- AdaBoost (classifier | regressor)10 min
- Gradient Boosting & HistGradientBoosting (sklearn)10 min
- XGBoost10 min
- LightGBM10 min
- CatBoost10 min
S09 · Build: Neural Networks for Tabular Data (glimpse)1 topics · 0.2 h
- MLPClassifier | MLPRegressor10 min
S10 · Build: Combining Models (ensembles & hybrids)3 topics · 0.5 h
- Voting and averaging10 min
- Stacking and blending10 min
- Hybrid models10 min
S11 · Build: Model Comparison, Error Analysis & Interpretation4 topics · 0.7 h
- Compare models fairly on the leaderboard10 min
- Error analysis10 min
- Interpretation: coefficients, permutation importance, PDP and SHAP10 min
- Fairness check of the final classifier10 min
S12 · Build: Model Improvement & Hyperparameter Tuning5 topics · 0.8 h
- Manual tuning and sensible search ranges10 min
- GridSearchCV and RandomizedSearchCV10 min
- Bayesian optimisation with Optuna10 min
- Single vs ensemble vs hybrid: final comparison10 min
- Leakage-safe final model selection and one-time test10 min
S13 · Build: Save, Pipeline & Local Deployment4 topics · 0.7 h
- Save and load models safely10 min
- End-to-end prediction function with input validation10 min
- Interactive demo with Gradio10 min
- From demo to API and monitoring (glimpse)10 min
S14 · CAPSTONE: Two End-to-End Supervised Projects2 topics · 0.8 h
- Deliver both projects end to end40 min
- Final conceptual questions and viva10 min
Unsupervised Learning by Building (one technique, two mini projects)30 topics · 4.9 h
U01 · Build: Foundations — Learning Without Labels (uses all U-projects)3 topics · 0.5 h
- What unsupervised learning is and how it differs from supervised10 min
- Types of unsupervised problems (map to this module)10 min
- The unsupervised workflow and preparing data for distances10 min
U02 · Build: Clustering — A: Customer Segmentation with RFM (Online Retail II) · B: Electricity Day Types9 topics · 1.3 h
- Build the features to cluster10 min
- K-means: build the core loop once, then use scikit-learn10 min
- Choose the number of clusters10 min
- Hierarchical clustering10 min
- DBSCAN and HDBSCAN10 min
- Gaussian Mixture Models10 min
- Compare algorithms, profile and name the segments10 min
- Assign new data and check stability10 min
- Practice (optional): German Credit personas revisited1 min
U03 · Build: Dimensionality Reduction — A: Customer Behaviour Map · B: Handwritten Digits Map5 topics · 0.8 h
- Why reduce dimensions + build the data10 min
- PCA: variance, components, loadings and reconstruction10 min
- t-SNE and UMAP maps10 min
- Reduce before clustering? (experiment)10 min
- Exercise: reduce, plot and explain10 min
U04 · Build: Anomaly Detection — A: Credit-Card Fraud Detection · B: Electricity Load Anomalies7 topics · 1.2 h
- Types of anomalies and the 'model normal' framing10 min
- Statistical baselines: z-score, IQR and contextual rules10 min
- Isolation Forest10 min
- Local Outlier Factor and One-Class SVM10 min
- Clustering-based detection10 min
- Evaluate: hidden labels (A) and injected anomalies (B)10 min
- Choose a method and design the alert10 min
U05 · Build: Association Rules — A: Market Basket Analysis (Online Retail II) · B: optional German Credit patterns3 topics · 0.5 h
- Baskets, support, confidence and lift10 min
- Frequent itemsets and rules with Apriori / FP-Growth10 min
- Turn rules into actions10 min
U06 · Build: Evaluation, Industry Projects & Presentation3 topics · 0.6 h
- Evaluate unsupervised results fairly10 min
- Optional industry projects (pick one)1 min
- CAPSTONE: present and defend25 min
Part 3 · Build a Full-Stack App with Claude Code (all settings, connectors and tools)8 topics · 1.7 h
3.1 Claude Code setup1 topics · 0.2 h
- Settings files and memory10 min
3.2 Connectors1 topics · 0.2 h
- Add MCP servers (GitHub, database, browser, docs)10 min
3.3 Customise1 topics · 0.3 h
- Slash commands, hooks, subagents, skills and plugins15 min
3.4 Run it everywhere1 topics · 0.2 h
- IDE, desktop/web, headless and GitHub Actions10 min
3.5 Project: AI expense tracker4 topics · 0.9 h
- Plan and set up the project10 min
- Back end, database and login15 min
- AI receipt reading and the front end15 min
- Test, review and deploy15 min
Part 4 · Build a Mobile App with Vibe Coding (Android and iOS)4 topics · 0.9 h
4.1 Mobile essentials1 topics · 0.2 h
4.2 Project: habit tracker with AI coach2 topics · 0.5 h
- Screens, navigation and storage15 min
- Notifications, camera and the AI coach15 min
4.3 Build and publish1 topics · 0.3 h
- Installable builds and app-store release15 min
Module 2 · Project 1: Facial Emotion Detection (image classification with CNN)
15 topics · 3 h
2.1 Understand the problem1 topics · 0.2 h
- Define the product and how we will judge it10 min
2.2 Collect and label the data2 topics · 0.3 h
- Get FER-2013 and explore it10 min
- Labelling your own images for classification10 min
2.3 Prepare the image data4 topics · 0.7 h
- Images as data: pixels, channels and size10 min
- Load from folders and split train / validation / test10 min
- Data augmentation10 min
- Handle class imbalance10 min
2.4 CNN concepts and first model3 topics · 0.7 h
- Convolution, filters and feature maps10 min
- Pooling and CNN architecture10 min
- Build and train a CNN from scratch20 min
2.5 Improve with transfer learning1 topics · 0.3 h
- Use a pretrained model20 min
2.6 Evaluate the model2 topics · 0.3 h
- Test-set score, per-emotion results and confusion matrix10 min
- Explain mistakes with Grad-CAM10 min
2.7 Deploy the model2 topics · 0.5 h
- Build the Gradio app (face detection + emotion)20 min
- Publish and document10 min
Module 3 · Project 2: Helmet Safety Detection (object detection with YOLO)13 topics · 2.7 h
3.1 Understand the problem1 topics · 0.2 h
- Detection vs classification and the success metric10 min
3.2 Collect and label the data2 topics · 0.5 h
- Collect images and write labelling rules10 min
- Draw bounding boxes (data labelling)20 min
3.3 Prepare the YOLO dataset2 topics · 0.5 h
- YOLO folder structure, label format and data.yaml20 min
- Split and check label quality10 min
3.4 Detection concepts2 topics · 0.3 h
- Bounding boxes, IoU, confidence and NMS10 min
- How YOLO works10 min
3.5 Train the detector1 topics · 0.3 h
- Baseline, then train YOLO15 min
3.6 Evaluate the detector2 topics · 0.3 h
- Precision, recall, mAP and choosing the threshold10 min
- Error analysis10 min
3.7 Deploy the detector2 topics · 0.4 h
- Gradio app for image, video and webcam15 min
- Export to ONNX and document10 min
3.8 Next steps (optional)1 topics · 0.2 h
- The same YOLO for segmentation and tracking10 min
Module 4 · Project 3: Electricity Demand Forecasting (time series with RNN/LSTM)9 topics · 1.7 h
4.1 Understand the problem2 topics · 0.3 h
- Time-series basics and the success metric10 min
- Baselines to beat10 min
4.2 Prepare the sequence data2 topics · 0.3 h
- Time split and scaling10 min
- Sliding windows and the 3D input shape10 min
4.3 RNN concepts2 topics · 0.3 h
- RNN: networks with memory10 min
- LSTM and GRU: long memory10 min
4.4 Build and train1 topics · 0.3 h
- Build the LSTM model and training loop in PyTorch15 min
4.5 Evaluate the model1 topics · 0.2 h
- Score on the test period and compare10 min
4.6 Deploy the model1 topics · 0.3 h
- Gradio forecasting app and README15 min
2.3 Tokenisation1 topics · 0.2 h
- Sub-word tokens, padding and attention masks10 min
2.4 Using pretrained models2 topics · 0.3 h
- Pretraining, fine-tuning and zero-shot10 min
- The Hugging Face toolkit10 min
2.5 Mini exercise1 topics · 0.2 h
- Five pipelines in ten minutes10 min
Module 3 · Project 1: Banking Customer Query Classifier (fine-tuning BERT)10 topics · 1.8 h
3.1 Understand the problem1 topics · 0.2 h
- Define the product and how we will judge it10 min
3.2 Data and labelling2 topics · 0.3 h
- Load and explore Banking7710 min
- Label your own text data10 min
3.3 Prepare the data1 topics · 0.2 h
- Splits, label IDs and tokenisation10 min
3.4 Baseline1 topics · 0.2 h
- TF-IDF + logistic regression10 min
3.5 Fine-tune the Transformer1 topics · 0.3 h
- Fine-tune DistilBERT with the Trainer20 min
3.6 Evaluate the model2 topics · 0.3 h
- Test score, per-intent results and confidence threshold10 min
- Error analysis10 min
3.7 Deploy the model1 topics · 0.2 h
- Push to the Hub and build the Gradio app10 min
3.8 Next step (optional)1 topics · 0.2 h
- Entity extraction (NER) with the same tools10 min
Module 4 · Project 2: Smart FAQ Search (semantic search with sentence embeddings)9 topics · 1.5 h
4.1 Understand the problem1 topics · 0.2 h
- Define the product and how we will judge it10 min
4.2 Build the knowledge base and test set2 topics · 0.3 h
- Create the FAQ10 min
- Create the labelled test set (relevance labels)10 min
4.3 Baseline1 topics · 0.2 h
- Keyword search (BM25)10 min
4.4 Sentence embeddings1 topics · 0.2 h
- Embeddings and cosine similarity10 min
4.5 Build the search1 topics · 0.2 h
- Vector index and search function10 min
4.6 Improve the search1 topics · 0.2 h
- Re-ranking and hybrid search10 min
4.7 Evaluate1 topics · 0.2 h
- Compare methods and set a 'no answer' threshold10 min
4.8 Deploy1 topics · 0.2 h
- Gradio FAQ search app10 min
Module 5 · Project 3: Customer Chat Summariser (fine-tuning T5 for text generation)9 topics · 1.7 h
5.1 Understand the problem1 topics · 0.2 h
- Define the product and what a good summary is10 min
5.2 Data and references2 topics · 0.3 h
- Load and explore the dataset10 min
- Write reference summaries and a rating guide10 min
5.3 Prepare the data1 topics · 0.2 h
- Input–output pairs for an encoder-decoder10 min
5.4 How text is generated1 topics · 0.2 h
- Decoding: greedy, beam search, sampling10 min
5.5 Baselines1 topics · 0.2 h
- Rule-based and zero-shot baselines10 min
5.6 Fine-tune the model1 topics · 0.3 h
- Fine-tune FLAN-T5 with the Seq2Seq Trainer20 min
5.7 Evaluate1 topics · 0.2 h
- ROUGE, human rating and fact check10 min
5.8 Deploy1 topics · 0.2 h
- Gradio summariser app and README10 min
15 min
Step-by-step reasoning15 minPrompt chaining and self-consistency15 min2.3 Prompts in code1 topics · 0.2 h
- Reusable prompt templates10 min
2.4 Mini exercise1 topics · 0.2 h
- Improve a prompt and measure it10 min
Module 3 · Project 1: Customer Support Email Assistant (prompting, structured output, tool calling)9 topics · 1.5 h
3.1 Understand the problem1 topics · 0.2 h
- Define the product and how we will judge it10 min
3.2 Data and labels1 topics · 0.2 h
- Build the golden test set10 min
3.3 Structured output2 topics · 0.3 h
- Get valid JSON and validate it10 min
- Improve extraction with prompt versions10 min
3.4 Tool calling1 topics · 0.2 h
- Let the model call your functions10 min
3.5 Draft the reply1 topics · 0.2 h
- Reply generation with rules10 min
3.6 Run it reliably1 topics · 0.2 h
- Streaming, batching, retries and rate limits10 min
3.7 Evaluate1 topics · 0.2 h
- Score accuracy, quality, cost and speed10 min
3.8 Deploy1 topics · 0.2 h
- Gradio app with human approval10 min
Module 4 · Project 2: Company Policy Q&A Assistant (RAG — Retrieval-Augmented Generation)13 topics · 2.3 h
4.1 Understand the problem1 topics · 0.2 h
- Why RAG and how we will judge it10 min
4.2 Data and test set1 topics · 0.2 h
- Collect documents and write the test questions10 min
4.3 Prepare the documents2 topics · 0.3 h
- Read PDFs and split into chunks10 min
- Embeddings and the vector database10 min
4.4 Generate the answer2 topics · 0.3 h
- Answer prompt with citations and 'I don't know'10 min
- Chat memory and context limits10 min
4.5 Use a framework1 topics · 0.2 h
- The same RAG pipeline in LangChain and LlamaIndex10 min
4.6 Evaluate1 topics · 0.2 h
- Retrieval score, correctness, faithfulness and refusals10 min
4.7 Improve2 topics · 0.5 h
- Query rewriting, chunk size, re-ranking and metadata filters20 min
- Glimpse: advanced RAG patterns10 min
4.8 Safety2 topics · 0.3 h
- Input guardrails: personal data and harmful requests10 min
- Prompt injection and data access10 min
4.9 Deploy1 topics · 0.2 h
- Gradio chat with sources and logging10 min
Module 5 · Project 3: Fine-Tune a Small Open LLM (LoRA / QLoRA)9 topics · 1.7 h
5.1 Understand the problem1 topics · 0.2 h
- When to fine-tune (and when not)10 min
5.2 Build the dataset2 topics · 0.3 h
- Instruction format and chat templates10 min
- Create labels with a bigger model and review them10 min
5.3 Fine-tuning concepts1 topics · 0.2 h
- LoRA and QLoRA in simple words10 min
5.4 Train1 topics · 0.3 h
- Fine-tune with TRL's SFTTrainer20 min
5.5 Align with preferences (optional)1 topics · 0.2 h
- Preference tuning with DPO10 min
5.6 Evaluate1 topics · 0.2 h
- Compare with the base model and the big model10 min
5.7 Deploy2 topics · 0.3 h
- Merge, share and run locally10 min
- Serving with vLLM and estimating cost10 min
Module 6 · Project 4: Marketing Image Generator (diffusion models)7 topics · 1.2 h
6.1 Understand the problem1 topics · 0.2 h
- How diffusion models create images10 min
6.2 Generate images1 topics · 0.2 h
- Prompts, steps, guidance and seed10 min
6.3 Edit images1 topics · 0.2 h
- Image-to-image and inpainting10 min
6.4 Brand style1 topics · 0.2 h
6.5 Search images with text (optional)1 topics · 0.2 h
- Image + text embeddings with CLIP10 min
6.6 Evaluate1 topics · 0.2 h
- Human rating and CLIP score10 min
6.7 Deploy and use responsibly1 topics · 0.2 h
- Gradio app and responsible-use rules10 min
2.3 Tools1 topics · 0.2 h
- Design and build the tools10 min
2.4 Build the agent2 topics · 0.3 h
- LangGraph basics: state, nodes and edges10 min
- System prompt and planning10 min
2.5 Memory1 topics · 0.3 h
- Conversation memory and long-term facts15 min
2.6 Human-in-the-loop1 topics · 0.3 h
- Approval before actions15 min
2.7 Evaluate1 topics · 0.3 h
- Task success, tool use, steps, cost — and traces15 min
2.8 Deploy1 topics · 0.3 h
- Streamlit chat app with approvals15 min
Module 3 · Project 2: Company Tools as an MCP Server (Model Context Protocol)8 topics · 1.8 h
3.1 Understand the problem1 topics · 0.2 h
- Why MCP and how we will judge it10 min
3.2 MCP concepts1 topics · 0.2 h
- Host, client, server; tools, resources, prompts10 min
3.3 Build the server1 topics · 0.3 h
- An MCP server in Python15 min
3.4 Test and use1 topics · 0.3 h
- Test with the Inspector and a desktop host15 min
3.5 Connect to your agent1 topics · 0.3 h
- Use MCP tools in LangGraph15 min
3.6 Remote server1 topics · 0.3 h
- Local vs remote servers and authentication15 min
3.7 Security1 topics · 0.3 h
- Tool poisoning, least privilege and approvals15 min
3.8 Ship1 topics · 0.2 h
Module 4 · Project 3: Agentic RAG Helpdesk7 topics · 1.5 h
4.1 Understand the problem1 topics · 0.2 h
- Where plain RAG fails and how we will judge10 min
4.2 Retrieval as a tool1 topics · 0.3 h
- Give the agent its search tools15 min
4.3 Corrective loop1 topics · 0.3 h
- Grade, rewrite and search again15 min
4.4 Self-check1 topics · 0.2 h
- Is the answer supported?10 min
4.5 Single vs multi-agent1 topics · 0.3 h
- When to split into several agents15 min
4.6 Glimpse1 topics · 0.2 h
- GraphRAG and knowledge graphs10 min
4.7 Evaluate and deploy1 topics · 0.3 h
- Compare with plain RAG and ship15 min
Module 5 · Project 4: Multi-Agent Customer Support System10 topics · 2.4 h
5.1 Understand the problem1 topics · 0.2 h
- When to use multiple agents and how we will judge it10 min
5.2 Patterns and design1 topics · 0.3 h
- Multi-agent patterns15 min
5.3 Test conversations1 topics · 0.3 h
- Write the test conversation set15 min
5.4 Build in LangGraph1 topics · 0.3 h
- Supervisor and specialist agents15 min
5.5 Build in CrewAI1 topics · 0.3 h
- The same idea as a crew (framework comparison)15 min
5.6 Coordination1 topics · 0.3 h
- Shared state and simple consensus15 min
5.7 Connect across systems1 topics · 0.3 h
- A2A: Agent-to-Agent protocol15 min
5.8 Evaluate and debug1 topics · 0.3 h
- Routing, resolution, failures and cost15 min
5.9 Security1 topics · 0.3 h
- Guardrails for a multi-agent system15 min
5.10 Deploy1 topics · 0.3 h
- Chat app and next steps15 min
Module 2 · Project 1: Inbound Receptionist Agent on Vapi (plug-and-play)10 topics · 1.7 h
2.1 Plan the agent2 topics · 0.3 h
- Define the goal, scope and success measures10 min
- Design the call flow and write the voice prompt10 min
2.2 Build on Vapi2 topics · 0.3 h
- Create the assistant: transcriber, model, voice10 min
- Knowledge base for FAQs10 min
2.3 Tools and webhooks1 topics · 0.2 h
- Booking tools via a webhook API10 min
2.4 Phone and transfers1 topics · 0.2 h
- Phone number, transfer to a human, ending the call10 min
2.5 Test1 topics · 0.2 h
- Scripted test calls and reading call logs10 min
2.6 After the call1 topics · 0.2 h
- Post-call summary, structured data and CRM10 min
2.7 Measure and launch1 topics · 0.2 h
- Metrics, cost and go-live10 min
2.8 Other platforms1 topics · 0.2 h
- Same agent on another platform (Retell) — optional10 min
Module 3 · Project 2: Outbound Calling Agent on Vapi (reminders and lead qualification)6 topics · 1 h
3.1 Plan the campaign1 topics · 0.2 h
- Use case, consent and calling rules10 min
3.2 Build the outbound assistant1 topics · 0.2 h
- Personalised prompt with variables10 min
3.3 Make calls1 topics · 0.2 h
- Start calls by API and run a campaign10 min
3.4 Handle real-world outcomes1 topics · 0.2 h
- Voicemail, no answer, busy and retries10 min
3.5 Multi-step calls1 topics · 0.2 h
- Several assistants in one call (squads / hand-offs)10 min
3.6 Results and metrics1 topics · 0.2 h
- Write back to the CRM and measure the campaign10 min
Module 4 · Project 3: Build Your Own Voice Agent with LiveKit (web + phone, inbound + outbound)10 topics · 1.7 h
4.1 LiveKit basics1 topics · 0.2 h
- Rooms, participants, tracks and agent workers10 min
4.2 First agent1 topics · 0.2 h
- A voice agent in about 30 lines10 min
4.3 Tools and state1 topics · 0.2 h
- Function tools and call state10 min
4.4 Hand-offs1 topics · 0.2 h
- Multiple agents in one call10 min
4.5 Knowledge and languages1 topics · 0.2 h
- RAG lookup and Hindi/Hinglish support10 min
4.6 Telephony1 topics · 0.2 h
- SIP trunk and inbound phone calls10 min
4.7 Telephony1 topics · 0.2 h
- Outbound calls and transfer to a human10 min
4.8 Speech-to-speech option1 topics · 0.2 h
- Use a realtime model and compare10 min
4.9 Measure1 topics · 0.2 h
- Recording, transcripts and latency metrics10 min
4.10 Test1 topics · 0.2 h
- Run the test scenarios and compare with Vapi10 min
Module 5 · Project 4: Production Voice-Agent Platform (your own 'mini-Vapi')12 topics · 2 h
5.1 Platform design1 topics · 0.2 h
- What is per client and what is shared10 min
5.2 Config-driven agents1 topics · 0.2 h
- Store agent settings in a database and load them per call10 min
5.3 Backend API1 topics · 0.2 h
- API for agents, numbers, campaigns and calls10 min
5.4 Campaign engine1 topics · 0.2 h
- Queue, concurrency, calling windows and retries10 min
5.5 Post-call pipeline1 topics · 0.2 h
- Transcript, summary, structured data, webhooks10 min
5.6 Dashboard1 topics · 0.2 h
5.7 Testing at scale1 topics · 0.2 h
- Simulated callers and regression tests10 min
5.8 Observability and reliability1 topics · 0.2 h
- Monitoring, alerts and provider fall-backs10 min
5.9 Deploy and scale1 topics · 0.2 h
- Deploy agent workers and the backend10 min
5.10 Security and compliance1 topics · 0.2 h
- Protect callers, clients and your platform10 min
5.11 Cost and pricing1 topics · 0.2 h
- Cost per minute and pricing your service10 min
5.12 Go live1 topics · 0.2 h
- Launch checklist and capstone demo10 min
Routes, Pydantic models and Swagger docs
10 min
2.3 Serve the ML model1 topics · 0.2 h
- Load once, predict, return explanations10 min
2.4 Serve the LLM app1 topics · 0.2 h
- Async calls, streaming, timeouts10 min
2.5 Errors and logs1 topics · 0.2 h
- Error handling and structured logging10 min
2.6 Security basics1 topics · 0.2 h
- API keys, CORS and rate limiting10 min
2.7 Tests1 topics · 0.2 h
- Automated tests with pytest10 min
2.8 Front end1 topics · 0.2 h
- Streamlit (or Gradio) app calling your API10 min
Module 3 · Project 2: Containerise, Deploy to the Cloud and Automate with CI/CD7 topics · 1.2 h
3.1 Docker basics1 topics · 0.2 h
- Images, containers and your first Dockerfile10 min
3.2 Several services1 topics · 0.2 h
- Docker Compose: API + front end + vector DB + cache10 min
3.3 Deploy (simple)1 topics · 0.2 h
- Deploy to a platform: Render or Railway10 min
3.4 Deploy (cloud)1 topics · 0.2 h
- Google Cloud Run with Artifact Registry and Secret Manager10 min
3.5 CI1 topics · 0.2 h
- GitHub Actions: test every pull request10 min
3.6 CD1 topics · 0.2 h
- Deploy automatically on merge10 min
3.7 Kubernetes (glimpse)1 topics · 0.2 h
- When and why Kubernetes10 min
Module 4 · Project 3: LLMOps and MLOps — Monitor, Evaluate and Improve the Live App10 topics · 1.7 h
4.1 Tracing1 topics · 0.2 h
- Trace every LLM request10 min
4.2 Prompt management1 topics · 0.2 h
- Version prompts outside code10 min
4.3 Evaluation gate1 topics · 0.2 h
- Evaluation in CI: block bad releases10 min
4.4 Online monitoring1 topics · 0.2 h
- Feedback, quality sampling, cost and latency10 min
4.5 Metrics and alerts1 topics · 0.2 h
- Prometheus and Grafana dashboards, alerts10 min
4.6 Cost and speed1 topics · 0.2 h
- Caching, model routing and fall-backs10 min
4.7 Guardrails in production1 topics · 0.2 h
- Input and output checks at the API10 min
4.8 MLOps for the ML model2 topics · 0.3 h
- Experiment tracking and model registry (MLflow)10 min
- Data and prediction drift, retraining10 min
4.9 Safe releases1 topics · 0.2 h
- A/B tests and canary roll-outs10 min
Module 5 · Project 4: Production Architecture for GenAI and Agent Systems (capstone)9 topics · 1.5 h
5.1 Architecture review1 topics · 0.2 h
- Design the system on one page10 min
5.2 Serving agents1 topics · 0.2 h
- Agents and chains behind a clean API10 min
5.3 Background jobs1 topics · 0.2 h
- Queues and workers for long tasks10 min
5.4 Self-hosted models (glimpse)1 topics · 0.2 h
- When to serve your own model10 min
5.5 Security and tenancy1 topics · 0.2 h
- Authentication, authorisation and multi-tenancy10 min
5.6 Load testing and scaling1 topics · 0.2 h
- Prove it handles the traffic10 min
5.7 Reliability1 topics · 0.2 h
- Timeouts, retries, health checks, graceful failure10 min
5.8 Governance1 topics · 0.2 h
- Audit logs, data retention and budgets10 min
5.9 Ship1 topics · 0.2 h
- Runbook, live deployment and showcase10 min
2.4 Run reliably1 topics · 0.3 h
- Test, handle errors, monitor and hand over15 min