GenAI, RAG, Agentic AI — and the engineering to ship them. A program that turns your students from people who have heard of LLMs into people who can build with them. Not another theory course; a build course.
Most students can define an LLM, name a few models, and describe RAG in a viva. Almost none of them can stand up a working RAG system, wire a tool-using agent, or ship something that survives contact with real input. That distance is exactly what employers are hiring across right now.
Industry has moved past analytics and classical ML into generative and agentic systems — and the roles opening up are for people who can build, integrate, evaluate, and operate these systems, not just talk about them. Most CS/IT curricula stop at the theory of machine learning, a stage or two behind.
This program closes that gap with a build-first approach. Students leave having architected an LLM application, grounded it with retrieval, given it tools, and shipped it with the guardrails that keep it from embarrassing them. The same engineering discipline, whatever domain they apply it to next.
Each stop produces working software, not slides. Concepts are introduced just-in-time, then immediately put to work at the keyboard. By the end, the pieces compose into one shipped application.
Stand up an LLM app · ground it with RAG · give it tools and make it an agent · integrate it with real systems · and ship it without it falling over.
Every topic blends four modes — concept lectures, guided builds, DIY builds, and take-home work — but the centre of gravity is the keyboard. Students spend most of their time building, with theory introduced exactly when they need it to take the next step.
The course is language-and-stack pragmatic: Python where it helps, but the focus is on composing modern AI building blocks — models, retrieval, agents, tools — into working systems, the way the industry actually does it today.
Engineering judgment runs through all of it: knowing how AI systems fail, how to evaluate them, and how to ship something you can stand behind. That discipline — not any single framework — is what makes a graduate employable in AI roles.
The program is project-based and assessed. Students don't submit a report — they demo a working application and defend the engineering decisions behind it. The certificate is co-brandable with your institution.
Five build stops, each producing working software. Guided builds in session, DIY builds as take-home.
The pieces come together into one project — an LLM application that retrieves, reasons, uses tools, and ships.
A project viva: demo the app, defend the engineering. Co-brandable certificate — a build, not a folder of slides.
The program flexes to your calendar — a concentrated on-campus block, or spread out weekly across a term. The content and the certification are identical; only the cadence changes. A short pre-read gets students to a common baseline before day one.
Why a builder, not a lecturer. This program is taught by someone who designs, ships, and governs AI systems in production — not someone teaching from a textbook. Students learn the engineering as it's actually practised, including the parts that only show up when a system meets real users and real data.
A build-first program, taught on your campus, ending in a shipped application and a real, defended credential. GenAI, RAG, Agentic AI — and the engineering to ship them.