Risk Academy · Programme outline
Machine Learning, Deep Learning, LLMs and AI Agents for Risk
Learning methods for risk, from validated tabular models to language models and tool-using agents, with evaluation and control built in.Acquis d'apprentissage
- Build, tune and validate a supervised model with leakage-safe splits and calibrated probabilities.
- Train a neural network on imagery or time series and evaluate it against a simple baseline.
- Design a retrieval-grounded LLM workflow with source citation and measured error rates.
- Specify an agent's tools, permissions and human approval points, and test its failure modes.
Existing course sources: geoai-for-risk, ai-emerging-technology-risk, model-risk-management.
New course outlines: Machine Learning for Risk; Deep Learning for Risk and Earth Observation; LLMs and AI Agents for Risk.
Modules, labs and assessments
- Supervised learning
Topics: splits, leakage, regularisation, calibration.
Lab: Predict a synthetic default label. Assessment: Graded model exercise. - Evaluation
Topics: metrics by cost, drift, subgroup performance, baselines.
Lab: Evaluate against a naive baseline. Assessment: Evaluation report. - Deep learning
Topics: neural networks, convolution, sequence models, transfer.
Lab: Segment water in satellite tiles. Assessment: Model report. - Large language models
Topics: transformers, prompting, retrieval, grounding, evaluation sets.
Lab: Build a cited answer pipeline over a document set. Assessment: Error audit. - AI agents
Topics: tool use, permissions, logging, human approval, failure modes.
Lab: Red-team a tool-using agent in a sandbox. Assessment: Safety case.
Capstone and credential
A documented model or agent for a stated risk task with a model card, an evaluation against a baseline and a list of known failures.
Completion earns a completion record and an achievement badge through the NXS Passport. It records that the learner finished the programme's assessed work. It does not certify competence and is not an accreditation.
Sources
- Hastie, T., Tibshirani, R., Friedman, J. (2009). The Elements of Statistical Learning, 2nd edition. Springer.
- Goodfellow, I., Bengio, Y., Courville, A. (2016). Deep Learning. MIT Press.
- Vaswani, A. et al. (2017). Attention Is All You Need. NeurIPS 30.
- NIST (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1.
- Mitchell, M. et al. (2019). Model Cards for Model Reporting. FAT* 2019.