Turn your engineers into the people who ship AI into production.
A structured cohort that takes strong engineers and gives them the second half of the job: finding the right problem, building within real constraints, and getting a system into production and into use.
Who this is for
Senior engineers with real production experience who are being asked to deliver AI outcomes and have never done it before. Not a beginner course. Participants should already be able to build software; what they are missing is the method for making AI work inside an organisation.
Typical participants: senior software engineers, data engineers, ML engineers, solutions architects, and engineering leads who will own an AI initiative.
The same method also prepares engineers for the adjacent roles companies are hiring for now: applied AI engineer, AI solutions engineer, AI architect, and GenAI engineer.
And if you would rather not train at all, we supply forward-deployed engineers directly, or hire them and employ them for you on the EOR model. FDE supply →
What they learn
The programme runs two tracks in parallel, because the role needs both and most engineers only have one.
| Track | Covers |
|---|---|
| Technical | Architecture for AI systems, evaluation, monitoring and observability, reliability and failure handling, deployment patterns, cost and latency optimisation, security in production |
| Consulting | Problem discovery, opportunity assessment, scoping and success measures, stakeholder alignment, delivery cadence, capability handover, saying no to the wrong project |
Structure
| Duration | 32 weeks, eight months |
| Weekly commitment | 14–18 hours |
| Format | Remote, with optional in-person sessions |
| Cohort size | Deliberately small |
| Projects shipped | 24 production projects, not exercises |
| Structure | Parallel technical and consulting tracks |
| Assessment | Capstone engagement simulation, reviewed by practising FDEs |
Learn, apply, build. Every module ends in something shipped, and every deliverable is judged against one question: would this survive inside a real organisation?
Every module ends in something built. The test applied to each deliverable is whether it would survive inside a real organisation reviewed code, real constraints, and a defensible design decision behind every choice.
Data used in labs
Labs run on synthetic or anonymised data by default. Where a cohort is customised to your stack, we work from examples you approve, and nothing leaves your control without your sign-off.
Certification
Participants who complete the programme and pass the capstone receive a Certificate .
Custom cohorts
A standard cohort teaches the method on neutral ground. A custom cohort teaches it on your architecture, your data estate and your domain, so what your engineers build during the programme is work you can actually use.
What gets customised
- Labs and case studies rebuilt around your stack and your domain
- Data examples drawn from your own estate, anonymised
- Up to 25% of modules swapped for ones your team needs more
- Capstone framed as a real problem from your operation
That scope is included in the cohort price. Deeper redesign new modules built from scratch for your context is scoped and priced separately as curriculum design.
How we scope one
- Scoping conversation. Your stack, your team's level, what you need them able to do afterwards.
- Capability assessment.
- Curriculum proposal. Written, module swaps named, price fixed.
- Schedule. Dates, cohort size, in-person elements if any.
Minimum cohort size for a custom run is 15 participants. Standard cohorts run first; custom cohorts dates can be discussed.
See when the next cohort starts.
Dates, format and seats are published on the cohort calendar. If the dates do not work, tell us the team and the timing you need.