01 / FOUNDATIONS
Computational Intelligence
Problem formulation, search, logic, probability, optimisation and computational approaches to intelligent behaviour.
DISCIPLINE / ARTIFICIAL INTELLIGENCE
Build intelligent systems. Understand the models.
AI at TUTOR is structured as a capability discipline: from computational foundations and data to machine learning, deep learning, generative models, intelligent agents and advanced AI systems. Learners move from using models to understanding, building, evaluating and architecting them.
Students presenting healthcare AI robots at a 2024 robotics event in Delhi · Wikimedia Commons · source below
01 / AI FIELDS
From models to intelligent systems.
The discipline separates core AI methods from the systems built around them. A learner can develop depth in one field while connecting it to the others.
02 / LEARNING
Machine Learning
Data, features, supervised and unsupervised learning, evaluation, generalisation and model selection.
03 / DEEP LEARNING
Neural Systems
Representation learning, neural architectures, training, optimisation and model evaluation.
04 / LANGUAGE
Language & Multimodal AI
Natural language, embeddings, retrieval, multimodal representations and model interaction.
05 / GENERATIVE
Generative AI
Generative models, prompting, retrieval-augmented systems, evaluation and controlled generation.
06 / SYSTEMS
Intelligent Systems
Agents, tools, perception, planning, memory, orchestration, deployment and human-system interaction.
02 / CAPABILITY ARCHITECTURE
Six levels. Increasing system depth.
The same progression used across TUTOR disciplines is adapted to AI. The capability changes at each level; the learner is not simply consuming progressively harder content.
LEVEL 01
Foundations
Understand data, algorithms, basic probability, computational thinking and what an AI model does.
LEVEL 02
Practice
Train and use classical ML models, prepare datasets, evaluate outputs and build small experiments.
LEVEL 03
Build
Develop end-to-end ML or deep-learning projects with data pipelines, experiments and reproducible evaluation.
LEVEL 04
Professional
Engineer production-oriented AI systems, model services, evaluation pipelines, observability and responsible deployment.
LEVEL 05
Architect
Design interconnected intelligent systems combining models, retrieval, tools, agents, data and software infrastructure.
LEVEL 06
Frontier
Conduct advanced experiments and research across foundation models, multimodal AI, agents, efficient learning and emerging methods.
03 / APPLICATION PATHWAYS
One discipline. Multiple scales of application.
AI capability can move from individual experimentation to complex intelligent systems. The scale does not replace the need for domain expertise and system constraints.
01 / SELF
AI Literacy & Tools
Use AI systems critically, formulate problems, evaluate outputs and build personal workflows.
02 / PROJECT
Applied AI
Build focused models, assistants, classifiers, recommenders or generative applications.
03 / ORGANISATION
AI Systems
Connect data, models, software, evaluation and users into operational systems.
04 / ENTERPRISE
AI Architecture
Design scalable model platforms, agentic workflows, governance and system interfaces.
05 / FRONTIER
Research & Creation
Investigate new methods, architectures, evaluation approaches and emerging intelligent-system paradigms.
04 / PROJECTS & EVIDENCE
Build evidence of actual capability.
AI learning should leave behind inspectable artefacts: datasets, experiments, models, evaluations, system traces, documentation and research outputs.
Model work
Projects focused on learning algorithms and model behaviour.
- Dataset preparation and analysis
- Baseline → experiment → evaluation
- Model comparison and error analysis
- Reproducible training workflows
- Model cards / technical documentation
System work
Projects focused on turning models into useful intelligent systems.
- Retrieval and knowledge-grounded applications
- Tool-using or agentic workflows
- Human-in-the-loop interfaces
- Evaluation and observability
- Deployment and system architecture
05 / DEEP LEARNING
Understand the model, not just the interface.
Neural networks provide a useful bridge between mathematical representation, training dynamics and modern intelligent systems.
From neurons to architectures.
Students can progress from understanding individual computational units and layers to studying architectures, training behaviour, representation learning and the evaluation of learned systems.
RepresentationOptimisationTrainingEvaluationGeneralisationArchitecture
06 / AI LIBRARY
Code, data, models, papers and experiments.
The library is organised by level and capability rather than as one undifferentiated list of AI resources.
TEXTS
Foundational Reading
Core mathematics, probability, algorithms, machine learning and AI concepts.
CODE
Frameworks & Notebooks
Programming environments, experiments, reference implementations and reproducible notebooks.
DATA
Datasets & Benchmarks
Datasets, evaluation sets, benchmark tasks and documentation for experimental work.
MODELS
Model Resources
Model documentation, architectures, checkpoints and evaluation material where licensing permits.
RESEARCH
Papers & Preprints
Research literature, technical reports, surveys and experiment-oriented reading.
SYSTEMS
Deployment & Evaluation
Serving, monitoring, testing, safety, security, reliability and system-level evaluation resources.
07 / RESPONSIBLE SYSTEMS
Capability includes knowing the system's limits.
Advanced AI work requires explicit evaluation of reliability, security, privacy, data quality, robustness and human interaction. These are engineering and research concerns, not decorative ethics sections.
EVALUATION
Measure
Define appropriate metrics, tests, baselines and failure analyses for the task.
ROBUSTNESS
Stress
Study edge cases, distribution shifts, adversarial conditions and unexpected inputs.
SECURITY
Protect
Consider data, model, tool, interface and infrastructure attack surfaces.
HUMAN
Interact
Design for human oversight, feedback, accessibility and understandable system behaviour.
DOMAIN
Constrain
Respect domain-specific safety, legal, regulatory and professional requirements where applicable.
08 / INTERDISCIPLINARY CONNECTIONS
AI is a connective discipline.
AI becomes substantially more useful when combined with a domain. The connection changes the system, the data, the evaluation and the capability required.
Computer Science
Algorithms, programming, distributed systems, software engineering, databases and computing infrastructure.
Business & Commerce
Decision systems, forecasting, optimisation, enterprise workflows, analytics and organisational intelligence.
Knowledge Systems
Language, knowledge representation, retrieval, textual corpora, semantic structures and digital humanities.
Engineering
Robotics, control, perception, simulation, scientific computing and specialised technical systems.
Physical & Health Education
Movement analysis, performance data, training systems and intelligent interfaces for physical practice.
Start with intelligence. Build systems.
Move from computational foundations to models, applications, architectures and frontier inquiry.
Artificial Intelligence · Courses
