A live cohort programme over six to eight weeks. Build a real AI product on AWS, ship it, then secure it properly — the way companies now need AI built.
Cloud security changed more in the last year than in the five before it.
If we haven’t met — I’m Soleyman. I went from £18K as an apprentice to £144K as a Senior Cloud Architect without a degree, and security was the part of that path that made the difference. I ran the first version of this programme in 2026.
Then the job changed. Companies started putting AI agents into their AWS accounts — things that read data, call APIs and make changes on their own. Almost every AI security incident this year came down to the same thing: an agent holding a permission it should never have had.
That is a cloud security problem, not a machine learning problem. Which means if you already know AWS, you are closer to this work than almost anyone else, and nobody has told you.
Most cloud security training is still teaching 2022.
You can find plenty of AWS security courses. You can find plenty of AI security talks. Almost nothing teaches you both together, and nothing lets you actually break an AI agent and then fix it.
So even after months of study, most engineers still hesitate on questions like these.
Can you explain how an AI agent actually gets compromised?
Can you write an IAM policy for an AI agent rather than a person?
Can you tell whether a Bedrock guardrail is doing anything useful?
Can you find an attack on an AI workload in CloudTrail?
If an agent gets hijacked, can you say how far it would get?
Can you explain the difference between a prompt-injection trick and a real security failure?
Can you show anyone proof that you have done this?
If those make you hesitate, that is not a knowledge gap you fix with more videos. You fix it by building it once, properly, in a real AWS account.
Five modules building AI on AWS. Three securing it.
Modules 1–3
You pick a real business problem, build an AI product for it with Claude Code, ship it to AWS and ground it in real business data. One product the whole way through, not a pile of disconnected labs.
Modules 4–5
You give it an agent that takes real actions, then make it reliable enough that other people could depend on it.
Modules 6–8
You lock down the account, attack your own agent five different ways, then fix it and run the same attacks again to prove they now fail.
Build AI on AWS
Module 1
First principles
Module 2
AI-native product
Module 3
Business data
Module 4
AI agents
Module 5
Production AI
Secure it
Module 6
Cloud security
Module 7
Agentic security
Module 8
Securing AI workloads
What's actually inside.
Eight modules, two live calls every week. Lessons and labs are yours to work through when it suits you.
01
Module 1
AI Engineering First Principles & Product Judgment
What an AI engineer actually does, and the judgment that separates them from someone who can prompt a coding tool. How to decide what’s worth building, what to leave out, and how to explain the trade-offs to an engineer and a business owner on the same day.
Inside this module
What an AI engineer actually does, and where the role is heading in 2027
Picking a real business problem worth solving, not a tutorial problem
Scoping the one product you’ll build, ship and secure for the rest of the programme
The real decisions behind StudyTech: what I built, what I cut, what I’d do simpler
Writing the one-page brief that explains it to a business owner
By the end of module 1
You have a scoped product, a real business problem behind it, and a brief you can defend.
02
Module 2
AI-Native Product Engineering
Building real software with AI, and actually understanding what you built. Claude Code as a pair rather than a crutch, and getting the product running on AWS rather than on your laptop.
03
Module 3
Business Data & AI Knowledge
Getting AI to give accurate answers about a real business, not confident made-up ones. Connecting the product to real data, and knowing when an answer can be trusted.
04
Module 4
AI Agents on AWS
Letting your product actually do things — call APIs, change records, take actions — with the right limits on what it’s allowed to do. What an agent is, how one works on AWS, and why the identity it runs as matters more than anything else.
05
Module 5
Production AI Engineering
Making it reliable enough that other people can depend on it. Evaluation, monitoring, cost and failure handling: the parts that separate a demo from a product.
06
Module 6
Cloud Security
Securing the AWS account your product lives in. How attackers get into AWS accounts, how identity works underneath every service, and protecting the data, secrets and network your product depends on.
07
Module 7
Agentic Security
Attacking the agent you built. Where agents are weak, how they get compromised, and why they’re different from every other workload you’ve secured.
08
Module 8
Securing AI Workloads on AWS
Bringing it all together. Designing your product to be secure from the start rather than locking it down afterwards, then running the same attacks again to prove they now fail.
You'll use Claude Code in every module.
Security engineers are expected to work with AI now, not just secure it. So rather than a separate module on it, you use Claude Code every week to do the work engineers used to do by hand.
ModuleWhat you use it for
1Turn a messy business problem into a one-page product brief
2Build and deploy the product, then explain every file it wrote
3Generate test questions and catch the answers your product made up
4Trace the agent’s permissions and map where it can be attacked
5Write the evals, and read the traces when something regresses
6Scan your account, read 30 days of CloudTrail and draft the least-privilege policies it actually needs
7Generate and run the attacks, and log which ones landed
8Build a security analyst that reviews your logs and flags what matters
One rule, every week
Claude drafts it. You check it.
Every submission includes what it got wrong. You can't spot a bad IAM policy you don't understand, so this keeps the learning in — and gives you the habit employers actually want: using AI and verifying it, rather than trusting it.
Everything you get.
The full curriculum — lessons and labs across all eight modules
Two live calls a week — a teaching session, then a hot seat where you bring your work and I pull it apart
The target AI agent — a deliberately insecure agent in Terraform, one command to deploy into your account, yours to keep
Both scoring scripts — run them on any AWS account and any agent, forever
The Claude Code pack — the prompts, pipeline hooks and analyst template used in the labs
The IAM audit checklist — for reviewing roles, policies and permissions on real environments
The breach analysis template — the framework used in module 6, reusable on any incident
The Cloud Security Engineer CV template — the structure that gets past a screener, with the projects you build here already slotted in
100 interview questions and answers — the questions cloud security engineers actually get asked, with the answers worked through
Private community and support — a coach on lab support so you’re not stuck at 11pm
Lifetime access — including every future version of the programme
Cloud security pays. AI security pays more.
Cloud security has been one of the better-paid specialisations in tech for years. What's changed is that companies now need people who can do it for AI workloads, and there are very few of them.
RoleExperienceTypical range
Cloud Engineer moving into security0–2 yrs security$110K–$140K
Cloud Security Engineer2–4 yrs$140K–$185K
Senior Cloud Security Engineer4–7 yrs$185K–$240K
Security Consultant or Architect7+ yrs$230K–$350K+
Ranges vary a lot by country, company and how you interview. No programme can guarantee you a salary, and I won't pretend otherwise. What this programme can do is make you a much stronger candidate with proof to back it up.
From the original Accelerator
From Tech Support to $200K+ Cloud Security Engineer at American Express.
Saleem used the Cloud Security Accelerator to build the skills, projects and confidence that landed him a Cloud Security Engineer role at American Express.
Saleem · Cloud Security Engineer · American Express
Is this programme right for you?
This is for you if
You want cloud security as your specialisation — the one corner of tech AI is creating work in rather than taking it away
You want to be AI-native: the person who can secure the agents everyone else is nervous about
You know your way around AWS, or you’re willing to get there
You’d rather have three real projects on GitHub than another certificate
You want proof you can talk through for forty minutes, not a certificate of completion
You can find eight hours a week for six to eight weeks
This is not for you if
You’re perfectly happy earning exactly what you earn today and wouldn’t change a thing
You’d rather not stand out in a job market that gets tighter every month
You want to watch videos and feel productive without building anything
You want exam prep with practice questions
You want a guaranteed job and a guaranteed salary before you’ve done any of the work
You’re not willing to troubleshoot when a lab breaks at 11pm
How this differs from everything else out there.
Other AI security trainingThis programme
FormatSlides, frameworks and conference talksYou deploy a real agent and attack it
CloudVendor-neutral, so nothing is hands-onBuilt entirely on AWS, in your own account
CoverageEither cloud security or AI security, rarely bothBoth, in the order they actually connect
ProofA certificate of completionFour repositories and a scored account
FeedbackNo way to know if you did it rightAutomatic scoring plus a weekly hot seat
AI toolingTreated as a separate subjectClaude Code used in every module
The target agent and both scoring scripts, yours to keep
Private community with coach lab support
Frequently asked questions.
Do I need any AI or machine learning background?
No. Modules 1 to 5 teach you to build with AI from first principles — what the model is, how it gets its facts, what an agent’s tools are, and which permission each AWS piece runs with. No maths, no model training. If you know your way around AWS, you have enough.
Can I do this alongside a full-time job?
Will this get me a job?
Does this prepare me for the AWS Security Specialty?
What happens when the programme ends?
What if I get stuck on a lab?
Your AWS account is locked down. The AI agent inside it isn't.
By the new year you'll either have built, shipped and secured a real AI product on AWS, or you'll have read about someone else doing it.
Cloud Engineer Academy