Roadmap · planned, not yet available
Where the curriculum goes next
These twelve advanced courses and eight maths studios are an architecture we are authoring and reviewing. None is published, enrollable or for sale yet, and we don't promise dates. Published courses are in the catalogue; new releases appear on updates.
ADV01 · Planned
Python, data and statistical reasoning for AI
Proposed: intermediate bridge, proposed 12–16 hours
5 planned modules
- M01: Python as a reproducible instrument
- M02: Data contracts and cleaning
- M03: Distributions and misleading summaries
- M04: Probability and uncertainty
- M05: Experiments and reliable conclusions
ADV02 · Planned
Machine learning that survives new data
Proposed: intermediate, proposed 12–16 hours
5 planned modules
- M01: Define prediction and establish a baseline
- M02: Splits, preprocessing and leakage
- M03: Regression and classification trade-offs
- M04: Metrics, calibration and decision thresholds
- M05: Unsupervised discovery and monitoring
ADV03 · Planned
Neural networks: build, train and diagnose
Proposed: intermediate/advanced, proposed 14–18 hours
5 planned modules
- M01: Tensors and forward computation
- M02: Losses, gradients and automatic differentiation
- M03: The training loop
- M04: Optimisation and generalisation
- M05: Architectures and model debugging
ADV04 · Planned
Transformers and adapting language models
Proposed: advanced, proposed 14–18 hours
5 planned modules
- M01: Tokenisation and representation
- M02: Attention and transformer blocks
- M03: Training objectives and generation
- M04: Adaptation choices
- M05: Efficiency and model provenance
ADV05 · Planned
Retrieval and grounded AI applications
Proposed: advanced application, proposed 10–14 hours
5 planned modules
- M01: Corpus and ingestion contracts
- M02: Chunking, indexing and retrieval
- M03: Grounded answer assembly
- M04: Retrieval attacks and permissions
- M05: Evaluation and corpus lifecycle
ADV06 · Planned
Agents, tools and MCP with controlled actions
Proposed: advanced, proposed 12–16 hours
5 planned modules
- M01: Workflow or agent?
- M02: Tool contracts
- M03: Planning, state and recovery
- M04: MCP integration
- M05: Permission boundaries and evaluation
ADV07 · Planned
Evaluations, experiments and evidence for AI quality
Proposed: advanced application, proposed 10–14 hours
5 planned modules
- M01: Define success and failure
- M02: Evaluation dataset design
- M03: Scoring methods and judge calibration
- M04: Experiments and uncertainty
- M05: Regression gates and live feedback
ADV08 · Planned
Multimodal AI: images, audio and documents
Proposed: advanced, proposed 12–16 hours
5 planned modules
- M01: Images, audio and timing as data
- M02: Vision and document extraction
- M03: Speech and structured notes
- M04: Cross-modal reasoning and generation
- M05: Evaluation, access and rights
ADV09 · Planned
Deployment, MLOps, reliability and cost
Proposed: advanced, proposed 14–18 hours
5 planned modules
- M01: Reproducible packages and contracts
- M02: Serving and failure management
- M03: Experiments, releases and rollback
- M04: Observability and drift
- M05: Cost and capacity decisions
ADV10 · Planned
AI privacy and security engineering
Proposed: advanced application, proposed 10–14 hours
5 planned modules
- M01: Data inventory and privacy design
- M02: Threat modelling AI boundaries
- M03: Defensive prompt-injection lab
- M04: Application and model supply-chain controls
- M05: Incident response and privacy review
ADV11 · Planned
AI product discovery, experience and business value
Proposed: advanced applied/nontechnical, proposed 8–12 hours
5 planned modules
- M01: Find the job, not the technology
- M02: Select the intervention
- M03: Design trust and control
- M04: Validate value and adoption
- M05: Price, cost and sustainable operation
ADV12 · Planned
Responsible AI governance and organisational adoption
Proposed: advanced applied/nontechnical, proposed 8–12 hours
5 planned modules
- M01: System context and accountability
- M02: Risk and control design
- M03: Evidence, fairness and transparency
- M04: Procurement and release review
- M05: Adoption, monitoring and change
Planned maths and statistics studios
- MAT01 Linear algebra, geometry and conditioning
- MAT02 Multivariable calculus and gradients
- MAT03 Optimisation and regularisation
- MAT04 Probability, likelihood and Bayesian reasoning
- MAT05 Statistical inference and experimental design
- MAT06 Causal questions and interventions
- MAT07 Time series, recommendation and decision quality
- MAT08 Sequential decisions and reinforcement learning