How long is the program?

Fourteen weeks of live weekly sessions, plus six months of continued access to the recordings, templates, and community.

Live online classes with a tutor, recorded for replay, supported by practical assignments and tutor feedback between sessions.

Will I build real projects?

Yes. Every curriculum level ends with an assignment, and the program finishes with a portfolio-ready project you present to your cohort.

What happens when I finish?

You finish with a production-grade capstone, a defensible architecture, and six months of continued access to the recordings, templates, and alumni community.

Can my employer pay for this?

Yes. Register with your work email and reply to your confirmation message, and we will send an invoice and a summary of the curriculum for your L&D team.

← All courses

Generative AI

AI Generalist Fellowship: Professional

Design, evaluate, secure, and deploy AI systems that hold up in production and survive a security review.

Duration7 classes live-online + 6 months recorded access FormatLive And Recorded LevelProfessional Next start12 Sep 2026

Program investment£750Enrolling

Course Overview

Build broad, practical confidence across the AI stack

The Professional level is about the parts of applied AI that only matter once something is live: evaluation harnesses, retrieval that holds up on real corpora, agent architectures with bounded autonomy, observability, cost control, threat modelling, and governance.

Ten curriculum levels move from architecture decisions through production retrieval and MCP, orchestration, integration engineering, deployment, evaluation, and security. The capstone is a production-grade AI solution with measured impact, a runbook, and an architecture you can defend. It suits engineers, technical leads, consultants, and operators who already build with AI and now own the consequences.

PrerequisiteExisting hands-on experience building with AI tools, APIs, or automations

Curriculum

A clear path from foundations to a finished AI solution

Each level combines practical concepts, guided tool use, and an assignment that moves your skills forward.

Level1

The Applied AI Systems Landscape

Frame the whole stack — models, context, tools, orchestration, and cost — so you can make architecture decisions rather than tool choices.

What you will learn

Compare hosted, open-weight, and small models against real constraints.
Choose between prompting, retrieval, fine-tuning, and workflow code.
Estimate latency, throughput, and cost before you commit to a design.

Build and practise

Write an architecture decision record for a real system in your organisation.
Model the unit economics of that system at ten and ten-thousand users.

Tools & platformsOpenRouterClaudeChatGPTArchitecture decision records

Level2

Advanced Prompting, Evaluation & Guardrails

Treat prompts as production code: versioned, tested against a dataset, and protected by guardrails you can prove work.

What you will learn

Build an evaluation set and score prompt changes instead of guessing.
Apply LLM-as-judge scoring and know where it misleads you.
Design guardrails for injection, refusal, and out-of-scope requests.

Build and practise

Build a fifty-case evaluation harness for one production-shaped task.
Break your own guardrails with an adversarial prompt set, then close the gaps.

Tools & platformsPromptfooLLM-as-judgePythonGit

Level3

Production Retrieval, Context & MCP

Build retrieval that holds up on real corpora, and expose your systems to models through the Model Context Protocol.

What you will learn

Tune chunking, hybrid search, and reranking against measured recall.
Keep a knowledge base fresh, permissioned, and auditable.
Expose internal tools and data to a model through an MCP server.

Build and practise

Build a retrieval pipeline and report precision and recall before and after tuning.
Ship an MCP server that exposes one internal capability safely.

Tools & platformspgvectorSupabaseModel Context ProtocolReranking models

Level4

Agentic Architectures & Orchestration

Design agent systems that stay predictable — clear tool contracts, bounded autonomy, and orchestration you can reason about.

What you will learn

Choose between single agent, multi-agent, and deterministic orchestration.
Define tool contracts, permissions, and failure behaviour precisely.
Bound autonomy with budgets, approvals, and stopping conditions.

Build and practise

Design and build an orchestrated multi-agent system for a real workload.
Run a failure drill and document how the system degraded and recovered.

Tools & platformsClaude Agent SDKn8nModel Context ProtocolPython

Level5

Generative Media at Production Scale

Run generative media as a pipeline: batch generation, brand consistency, review gates, and rights you can defend.

What you will learn

Automate batch generation with consistent style and quality control.
Build a human review gate into a media pipeline.
Handle licensing, provenance, and disclosure responsibly.

Build and practise

Build a pipeline that generates, reviews, and publishes a media set automatically.
Write the provenance and disclosure policy that pipeline operates under.

Tools & platformsComfyUIRunwayElevenLabsn8nCloud storage

Level6

Automation & Systems Integration Engineering

Integrate AI with the systems a business actually runs on, including the queues, retries, and idempotency that keeps it honest.

What you will learn

Design idempotent, replayable jobs that survive partial failure.
Integrate with authenticated third-party APIs and webhooks safely.
Handle rate limits, backpressure, and long-running work.

Build and practise

Ship an integration that processes a real queue with retries and dead-lettering.
Prove it is idempotent by replaying a full day of traffic.

Tools & platformsn8nWebhooksMessage queuesREST and GraphQL APIs

Level7

Designing and Deploying AI Applications

Take an AI feature from prototype to something a team can deploy, support, and change without fear.

What you will learn

Design interfaces that set honest expectations about AI output.
Ship streaming, caching, and graceful degradation for model calls.
Deploy with environment separation, secrets handling, and rollback.

Build and practise

Deploy an AI application with streaming, caching, and a documented rollback plan.
Write the runbook a colleague would use to support it at 2am.

Tools & platformsNext.jsSupabaseVercelClaude API

Level8

Evaluation, Observability & Cost Control

Know what your AI system is doing in production, what it costs, and whether last week's change made it better or worse.

What you will learn

Instrument traces, token usage, latency, and quality signals.
Detect regressions with continuous evaluation on live traffic samples.
Cut cost with caching, routing, and right-sizing without losing quality.

Build and practise

Instrument a live system and build the dashboard you would actually check.
Reduce the cost of one workload by half and prove quality held.

Tools & platformsPrompt cachingModel routingTracing toolsDashboards

Level9

Security, Governance & Responsible Deployment

Deploy AI in a way that survives a security review, a privacy review, and a difficult question from a regulator or a customer.

What you will learn

Threat-model prompt injection, data exfiltration, and tool abuse.
Apply data minimisation, retention, and residency to AI systems.
Write the policy, audit trail, and human oversight a deployment needs.

Build and practise

Produce a threat model and mitigation plan for your capstone system.
Draft the AI usage policy and audit trail for one real team.

Tools & platformsOWASP LLM Top 10Audit loggingAccess controls

Level10

Capstone: Production AI Solution & Demo Day

Deliver a production-grade AI system with evaluation, observability, documentation, and a defensible architecture.

What you will learn

Deliver an AI system that is evaluated, observable, and documented.
Defend your architecture decisions and the trade-offs behind them.
Present measurable impact rather than a feature demonstration.

Build and practise

Ship a production-grade AI solution and present it at the professional demo day.

Tools & platformsLearner-selected AI stack

Schedule

Choose your cohort

Published dates and sessions update automatically as the Academy opens new cohorts.

Enrolling

September 2026 Cohort

Fourteen live Saturday classes from 12 September to 12 December 2026, plus architecture clinics and capstone review.

Starts: 12 Sep 2026
Format: Live And Recorded
Timezone: Europe/London

Published sessions

12 Sep 2026Cohort kickoff & solution scoping

10 Oct 2026Architecture review clinic

14 Nov 2026Evaluation & observability workshop

12 Dec 2026Production demo day

Course Tutors

Learn with active AI practitioners

Matthew Makinde

Course Tutor

Practical guidance that helps learners understand AI and put it to work.

Maro Orode

Course Tutor

Hands-on support for turning AI ideas into useful products and workflows.

Blessed Oigbochie

Course Tutor

Project-led teaching focused on confident, practical use of modern AI tools.

Godwin

Course Tutor

Learner-focused instruction across creative AI tools and real-world applications.

Take the next step

Register and secure your place

Complete your details, then continue to Stripe’s secure checkout to pay £750.

✓ Secure payment powered by Stripe✓ Your place is confirmed after successful payment✓ Your details are only used for this registration

Course FAQs

Before you apply

Practitioners who already build with AI and are accountable for what they ship — engineers, technical leads, consultants, and senior operators. You should arrive with hands-on experience of AI tools, APIs, or automation platforms.