AI Development Courses Built Around What the Work Actually Requires

A two-day session across one weekend. Takes a small tabular dataset from raw file to a trained, evaluated, and documented model. Covers exploration, leakage, baselines, gradient-boosted modelling, honest evaluation, and writing the model card. Runs in a provided sandbox — participants need working Python and a laptop.

This workshop teaches one path end to end. It does not cover deep learning, neural architectures, or production deployment. Those topics appear in the longer courses.

Thirteen weeks covering Markov decision processes, dynamic programming, temporal-difference methods, policy gradients, actor-critic architectures, and exploration strategies. Includes a module on evaluation under stochasticity and documented limitations of common benchmarks. Prerequisites: a foundations course or equivalent, plus comfort with probability. Five contact hours weekly plus ten hours of laboratory work.

Assessment is by three implementation exercises and a final reproduction study with a written account of what did not replicate and why.

A twelve-month retainer for organisations building internal AI engineering capability. Includes quarterly training blocks for cohorts of up to twenty staff, monthly instructor office hours, a tailored and quarterly-refreshed curriculum, code review clinics against internal repositories, a documentation and evaluation template library, and an annual capability report.

This retainer covers teaching, engineering practice, and process design. It carries no responsibility for production systems and makes no assessment of regulated obligations, which remain with the organisation's own qualified advisers.