The module that reframes the whole programme. It opens on data-driven decision-making and the full stack of what the words now mean — analytics, ML, deep learning, generative AI and agentic AI — including where AI-assisted decisions break down. Learners then practise business problem framing: stakeholders, KPIs, baselines, decision points, analytics maturity, value hypothesis and success criteria. From there into descriptive and diagnostic analytics, dashboards, exploratory insight and root-cause thinking; then data storytelling for executives — narrative structure, audience analysis, chart selection, annotation and insight-to-action communication. It closes on responsible analytics: privacy, fairness, explainability, AI risk, governance and business impact measurement.

Tools / Platforms: Excel / Google Sheets, Power BI, Tableau Public, Canva / PowerPoint, ChatGPT / Copilot (with disclosure)

Portfolio artefact: A CXO-ready analytics story for a real business problem — KPIs, baseline, dashboard sketch, expected business impact and governance risks.

Week one, you are not learning Python. You are learning how to walk into a CXO’s room and make a number mean something.

The engineering floor of the programme, and deliberately assumption-free — no prior programming is required. Learners build Python essentials from environment setup through data types, control structures, functions, modules, exceptions and code quality; then data handling with NumPy and pandas across indexing, filtering, grouping, reshaping, merging, dates, strings and file formats. The SQL block goes well past SELECT — relational modelling, DDL/DML, joins, subqueries, CTEs, aggregation, window functions and optimisation basics. It closes on the work that separates an analyst from a hobbyist: ingestion and integration across Excel, CSV, JSON, databases and APIs, then data quality and pipelines — missing data, duplicates, outliers, validation rules, logging, reproducibility, Git and documentation.

Tools / Platforms: Python 3.x, Jupyter / Colab / VS Code, NumPy, pandas, matplotlib, SQLite / MySQL / PostgreSQL, Git and GitHub

Portfolio artefact: An end-to-end ETL pipeline for a business dataset — SQL extraction, Python transformation, data quality checks and an analysis-ready output table.

No prior programming is mandatory. The prerequisite is literally ‘comfort with spreadsheets’. Say that on the page — it is the objection that kills this programme.

Statistics taught as a decision tool, not a syllabus. Learners cover descriptive statistics, probability, distributions, sampling, the central limit theorem and how to reason about uncertainty; then inference — estimation, confidence intervals, p-values, effect sizes and the interpretation errors that appear in real boardrooms. Hypothesis testing and regression follow: t-tests, chi-square, ANOVA, correlation, linear and logistic regression, and diagnostics. The genuinely new half is experimentation and causal thinking — A/B testing, control groups, quasi-experiments, bias, confounding, attribution and uplift measurement — closing on decision science: decision trees, risk, expected value, sensitivity analysis and optimisation basics.

Tools / Platforms: Excel / Google Sheets, Python (scipy, statsmodels, pandas), Jupyter / Colab, Power BI

Portfolio artefact: A designed and analysed A/B test or controlled business experiment — hypothesis, sample logic, metrics, statistical test and a decision recommendation.

Causal inference and A/B testing are now core. Most analytics Masters still stop at correlation and hope nobody notices.

ML anchored to use-case fit and business value rather than leaderboard scores. The module covers supervised, unsupervised and semi-supervised learning, the ML lifecycle, data leakage and train-test-validation discipline; then regression and classification across linear and logistic regression, decision trees, random forest, gradient boosting, k-NN, SVM and model selection. Feature engineering and preprocessing follow — encoding, scaling, missing values, imbalance handling, pipelines and cross-validation — then clustering and segmentation with k-means, hierarchical clustering, DBSCAN and PCA. It ends on evaluation and interpretation: metrics, confusion matrix, ROC-AUC, F1, RMSE, feature importance, SHAP and, always, business impact.

Tools / Platforms: Python, scikit-learn, pandas, NumPy, matplotlib, Jupyter / Colab, SHAP

Portfolio artefact: A business ML model for churn, lead scoring, demand prediction or segmentation — with baseline comparison and a business recommendation.

Every model ends with a business recommendation. A model without one does not pass.

Deep learning grounded in business problems and deployment reality. Learners cover foundations — neurons, layers, activations, losses, optimisation, backpropagation, regularisation and the training workflow; then feed-forward networks for tabular business problems with hyperparameter tuning, overfitting control, batch normalisation and dropout. CNNs and computer vision follow, covering image classification, object detection concepts, transfer learning and visual inspection use cases; then sequence and representation learning across RNNs, LSTM/GRU, attention, embeddings and time-series and text applications. The final unit is the one most courses skip: deployment considerations — GPUs, model compression, latency, monitoring, explainability, responsible AI and actual business adoption.

Tools / Platforms: TensorFlow / Keras, PyTorch, Hugging Face Transformers, Colab GPU, Weights & Biases / TensorBoard

Portfolio artefact: A deep learning prototype for visual inspection, demand forecasting, document classification or sentiment analysis — with metrics and deployment considerations.

Deep learning moves to Semester II — after Python, statistics and classical ML are in place. The old sequencing taught it before learners could carry it.

From tokenisation to production-grade retrieval. Learners cover NLP foundations — tokenisation, stemming and lemmatisation, n-grams, TF-IDF, embeddings and text classification; then transformers and LLMs, covering attention, pre-training, fine-tuning concepts, instruction tuning, context windows and model selection. The prompt engineering block treats prompting as engineering, not folklore: role prompting, few-shot design, structured outputs, tool use and prompt evaluation. Then a full RAG build — document loading, chunking, embeddings, vector databases, retrieval, re-ranking, citations and knowledge-grounded QA. It closes on LLM evaluation and governance: hallucination, safety, privacy, bias, guardrails, monitoring, cost management and responsible deployment.

Tools / Platforms: Python, spaCy / NLTK, Hugging Face Transformers, LangChain / LlamaIndex, FAISS / Chroma / Azure AI Search, OpenAI / Azure OpenAI

Portfolio artefact: A RAG-based business knowledge assistant or customer support assistant — with an evaluation dataset, prompt strategy and risk controls.

Cost management and hallucination control are taught as first-class topics. That is what separates a demo from something an enterprise will actually run.

The differentiator. Learners cover agentic AI foundations — agents, tools, planning, memory, autonomy, multi-agent systems, orchestration patterns and enterprise use cases; then business process discovery: workflow mapping, automation feasibility, human-in-the-loop design, KPIs and the transformation value case. The build block covers agent design and implementation — prompt workflows, tool calling, API integration, documents, databases, actions and orchestration frameworks. Then testing and evaluation on the dimensions that decide whether an agent survives procurement: task success, reliability, latency, cost, hallucination, regression tests, red-teaming and user acceptance. It closes on governance and adoption — guardrails, audit logs, security, privacy, change management, operating model and ROI measurement.

Tools / Platforms: LangChain / LangGraph, Microsoft Semantic Kernel, LlamaIndex, Azure AI Foundry / OpenAI tools, Postman, GitHub

Portfolio artefact: A working agentic workflow for sales enablement, HR support, finance operations, admission counselling, service desk or compliance review — including ROI and risk controls.

A full 5-credit core module on agent design, orchestration, guardrails and ROI. Name one other M.Sc. in Business Analytics in India that has this.

A demo-ready individual implementation with a full-length report, validated by an industry mentor and defended at viva. Learners work through problem scoping and proposal — business context, objectives, stakeholders, data sources, feasibility and mentor approval; then literature review and methodology with 15+ references, gap identification and a CRISP-DM or technical implementation methodology. Then design and build across data understanding, preparation, modelling, dashboards, AI workflows or APIs; then testing and evaluation with model metrics, dashboard validation, benchmark comparison, business impact and risk analysis. It ends with documentation, a GitHub repository, a demo walkthrough and a panel defence.

Tools / Platforms: GitHub, Jupyter / VS Code / Colab, Power BI / Tableau, Python AI/ML libraries, Draw.io / Lucidchart, MS Teams for mentor reviews

Portfolio artefact: An individual capstone with working demo, report, code repository, presentation and viva.

10 credits · 300 hours · one industry mentor · one panel.

The whole commercial funnel, end to end. Learners cover customer data and KPIs — funnels, cohorts, acquisition, activation, retention, revenue, churn, NPS and journey analytics; then segmentation and customer value through RFM, CLV, persona analytics, clustering, cohort analysis and targeting strategy. The predictive block builds churn, lead scoring, propensity, cross-sell and upsell, recommendation systems and next-best-action. Marketing effectiveness follows: campaign analytics, attribution, uplift modelling, A/B testing, experimentation and marketing mix basics. It closes on revenue analytics — pricing, demand, pipeline analytics, forecasting, executive reporting and action planning.

Python (pandas, scikit-learn), Power BI / Tableau, Excel, SQL, Google Analytics and CRM Analytics

Portfolio artefact: A customer or revenue analytics solution that identifies growth levers and recommends measurable interventions.

Uplift modelling and attribution — the two things every growth team argues about and almost no graduate can actually do.

Three enterprise functions in one module, taught as one decision problem. Learners cover finance analytics foundations — financial KPIs, revenue, cost, margin, cash flow, profitability, budgeting, variance and scenario analytics; then forecasting through time-series decomposition, ARIMA/ETS concepts, regression forecasting, demand planning and forecast accuracy metrics. Risk analytics follows: credit, fraud and operational risk, anomaly detection, early warning indicators, compliance and model risk. Then operations and supply chain analytics — inventory, capacity, queues, scheduling, routing and logistics, quality and process mining. It closes on optimisation and decision support: linear programming, simulation, what-if analysis, dashboards and executive recommendations.

Tools / Platforms: Python (statsmodels, prophet / skforecast, scikit-learn), Excel Solver, Power BI, OR-Tools / PuLP, SQL

Portfolio artefact: A finance, risk or operations analytics prototype with forecast, risk or optimisation output and a decision recommendation.

Two modules became one — but 8 credits became 5. Address that on the page before a competitor does. See Section 7, Flag 4.

The module that turns an analyst into someone who gets invited to the strategy meeting. It covers consulting fundamentals — discovery, stakeholder interviews, problem framing, hypothesis trees, issue trees, MECE thinking and executive communication; then enterprise AI strategy: AI maturity, portfolio prioritisation, operating model, capability building, data and platform readiness, and adoption roadmap. Responsible AI governance follows — ethics, explainability, privacy, bias, DPDP and GDPR awareness, model risk, audit trails and compliance. Then value measurement and ROI: baselines, counterfactuals, A/B tests, benefit tracking, productivity metrics, cost-to-serve and risk reduction. It closes on transformation execution — change management, governance committees, vendor evaluation, AI CoE, policy and executive reporting.

Frameworks: NIST AI Risk Management Framework, OECD AI Principles, India DPDP Act, GDPR, Microsoft Responsible AI

Tools / Platforms: PowerPoint / Canva, Miro / FigJam, Excel for ROI models, NIST AI RMF templates

Portfolio artefact: An AI transformation consulting proposal for a real organisation — roadmap, governance model, ROI logic and implementation plan.

Consulting plus governance plus ROI in one module. This is the CXO track, and it is where the 39 LPA outcomes actually come from.

The advanced build — higher maturity, stronger methodology, demonstrable business or technical impact. It may extend Capstone I or address a new problem. Learners work through project selection and novelty with mentor validation and publication fit; then advanced methodology across data, architecture, modelling, experimentation, validation and governance planning; then implementation and deployment readiness — code quality, pipelines, dashboards, agents, APIs, cloud services and user testing. Evaluation covers technical metrics, baseline comparison, ROI, risk controls, limitations and reproducibility. It ends with report, demo, viva and — critically — conversion into a journal paper outline and submission plan that Capstone III then executes.

Tools / Platforms: GitHub, Python / SQL / Power BI or the AI stack as required, cloud sandbox, Mendeley / Zotero, Draw.io / Lucidchart

Portfolio artefact: An advanced individual capstone with working demo, report, code repository, presentation, viva and manuscript outline.

Capstone II moves to Semester III — so the manuscript outline exists before Semester IV starts. That is why publication rates hold.

Aligned to Microsoft Exam AI-103: Developing AI Apps and Agents on Azure. Learners cover Azure AI solution architecture — Microsoft Foundry, model selection, resource setup, identity, access and environment configuration; then generative AI app development with prompts, system messages, structured outputs, content safety, multimodal inputs and the Python SDK. Knowledge and RAG solutions follow: Azure AI Search, embeddings, indexes, chunking, grounding, citations, evaluation and enterprise data integration. Then agent development — tools, actions, function calling, orchestration, memory, connected APIs, task workflows and user interaction design. It closes on evaluation, deployment and operations: responsible AI, security, monitoring, cost, testing, deployment options and exam readiness.

Aligned Certification: Microsoft AI-103 — Azure AI Apps and Agents Developer Associate

Tools / Platforms: Azure AI Foundry, Azure AI Search, Azure OpenAI / Azure AI services, Python SDKs, VS Code, GitHub

Portfolio artefact: A deployed Azure AI app or agent aligned to AI-103 skills — code, evaluation, security and an exam-prep checklist.

★ AI-103 is the current-generation exam. The live page still advertises AI-102. Fixing that is the single highest-value change on this page.

Aligned to Microsoft Exam AI-300: Operationalizing Machine Learning and Generative AI Solutions. Learners cover AI operations foundations — MLOps, GenAIOps, Azure Machine Learning, Microsoft Foundry and lifecycle concepts; then infrastructure and experimentation across workspaces, compute, environments, data assets, feature management, experiment tracking and reproducibility. Pipelines and automation follow: training pipelines, CI/CD with GitHub Actions or Azure DevOps, model registry, deployment strategies and endpoint management. Monitoring and governance covers data and model drift, performance monitoring, alerts, lineage, responsible AI dashboards, security and access control. It closes on GenAIOps — prompt flows, prompt and version management, evaluation datasets, safety metrics, monitoring and cost optimisation.

Aligned Certification: Microsoft AI-300 — Machine Learning Operations Engineer Associate

Tools / Platforms: Azure Machine Learning, Azure AI Foundry, GitHub Actions / Azure DevOps, MLflow, Docker, Azure Monitor

Portfolio artefact: An Azure MLOps/GenAIOps pipeline with automated training and deployment, a monitoring plan, governance artefacts and an AI-300 readiness checklist.

★ GenAIOps — prompt flows, evaluation datasets, safety metrics, cost optimisation. This did not exist as a job two years ago. Now it is a certification.

Converts Capstone I, Capstone II or a synthesis of both into a Scopus-indexed journal submission. Learners cover research contribution identification — gap, novelty, problem significance, methodology fit and publication positioning; then manuscript structure from title and abstract through literature review, methodology, results, discussion and references. Academic writing and evidence follows: tables, figures, citations, reproducibility, limitations and ethical disclosure. The module then covers journal selection and publication ethics — Scopus verification, Q-rating, predatory journal avoidance, authorship, plagiarism and data integrity — and closes on the submission and review process: cover letter, author guidelines, response to reviewers and revision planning.

Tools / Platforms: Mendeley / Zotero, Overleaf / MS Word, Turnitin, Scopus Sources, Google Scholar

Portfolio artefact: A manuscript submitted to a Scopus-indexed journal, with proof of submission or acceptance.

Pull-quote / hook line:  Predatory journal avoidance and Scopus Q-rating verification are taught explicitly. That is a quality signal worth saying out loud.