THE ACADEMY
The AI shift moved faster than anyone's playbook.
Ryvane Academy is built for three people the AI shift caught off guard. The newcomer to security, watching the feeds fill with talk of automated pentesting and layoffs, unsure the field is still worth entering. The working security professional who can feel how fast things are moving, but has no clear place to start with the AI-integrated apps now landing on their desk. And the developer shipping AI features under pressure to compete, where the security of what they just integrated quietly becomes tomorrow's problem. The academy is our answer to all three: a structured path to get AI-ready, and proof that AI-assisted security is something you can do from day one.
CURRICULUM PREVIEW
Courses to make yourself AI-ready.
AI/ML Foundations for Security Professionals
The mental model. How models learn, where they fail, and the vocabulary every later course assumes. No prior ML required.
10 modules40 MCQs6 challenges
Transformer Architecture
Embeddings, attention, residual streams, decoding. A defender's tour of the architecture that ships almost every modern AI system.
12 modules55 MCQs8 challenges
Machine Learning for Security Professionals
Classifiers, embeddings, training loops, and evaluation. Enough hands-on ML to read papers, run experiments, and call out unsafe shortcuts.
11 modules50 MCQs9 challenges
MCP Security
A dedicated course on the Model Context Protocol surface. Server design, capability scoping, tool-call boundaries, and the failure modes attackers are mapping today.
9 modules42 MCQs10 challenges
LLM Security & Prompt Injection
Direct and indirect injection, jailbreaks, exfiltration chains, and defenses that survive real users. Mapped to OWASP LLM Top 10 and MITRE ATLAS.
10 modules48 MCQs14 challenges
Agentic AI Security
Tool-use, planning, memory, and multi-agent traces. The new attack surface the moment an LLM gets the ability to act, not just answer.
9 modules44 MCQs11 challenges
WHY FREE
Honest about how it gets made.
Let's be honest. In the current LLM era, creating content that used to take a long time is no longer hard, especially the text. Our operators are, and will continue to be, using LLMs to help draft these courses.
But the manual work is real: reviewing every claim, rewriting passages in the way we'd actually explain them, designing the interactive components & challenges, keeping each lesson grounded in the audit work we're shipping. Putting in that effort is what helps us become better AI security experts ourselves, and it's what makes the academy actually useful for you. :)
INSIDE EACH COURSE
Some of the features that make learning actually engaging.
A glimpse of what's inside the platform. These are just a few of the interactive components built into every course, not the full list, but enough to give you a feel for how learning here actually plays.
Certificate of Completion
This is to certify that
Alex Morgan
has successfully completed
LLM Security & Prompt Injection
01 · MCQ CHECKPOINT
Read the concept. Prove it stuck.
Every lesson ends in a short, scored MCQ. We tell you which option is right, which traps the field commonly falls for, and why. Wrong answers come with a one-paragraph deep-dive, not a red x.
- 3-5 questions per lesson
- Distractors written from real audit findings
- Explanations on every option, not just the correct one
02 · ANNOTATED DIAGRAM
The architecture, stepped through.
Transformers, agent loops, MCP traces. Concepts come with diagrams you advance one step at a time, with each part of the system labeled as the data flows through it. Built for security readers, not ML PhDs.
- Click to advance, no autoplay
- Every callout cross-links to the relevant lesson
- Diagrams downloadable as SVG
03 · CODE CHALLENGE
Write the fix. Run it. Watch it land.
Every offensive lesson has a defensive twin. Patch the guardrail, harden the prompt template, or tighten the MCP capability scope, then run the challenge and watch your patch hold against the original attack.
- Browser-based editor, no setup
- Hidden test suite, instant verdict
- Reveal the canonical answer after you pass
$ ryv challenge run 07
passed
04 · HANDS-ON LAB
A real terminal. A real system to break.
Selected lessons unlock a full lab: a live, sandboxed system with a brief, a hint, and a goal. Step through it on your own clock, save your run, and come back to it as evidence the work was done.
- Persistent sandboxed environment
- Hints unlock progressively
- Reproducible from your saved run
LAB BRIEF
Find a tool-use call that the model can be coaxed into running from indirect content alone. Capture the trace.
Frequently Asked Questions
Yes. The operators building these courses already have full-time roles, and we genuinely enjoy creating them. At this stage it does not take much effort to keep them going, so there is no reason to put them behind a paywall.
Security professionals who suddenly need to defend AI systems, and ML engineers who need to ship them safely. Solid software fundamentals are assumed. No prior ML or AI background is required, the Foundations course exists for exactly that reason.
You get a verifiable completion record tied to the challenges you cleared, not a certificate-mill PDF. The goal is to make it easy for a hiring team or an internal review board to confirm what you actually solved, not what you watched.
Every course is a sequence of focused, text-based lessons, with MCQs and challenges in every section to refresh and test your learning, animated visuals to explain certain concepts, and most modules end in a practical hands-on challenge. You move at your own pace, in any order, with progress saved across devices.
Fully on-demand. The academy is not a cohort, there is nothing to sign up for at a specific time. Open a course, read a lesson, do the checkpoint, come back tomorrow.
Foundational development, computer, and cyber security skills.
No. Challenges run in a sandboxed environment in your browser. Editor, terminal, target system, and grader all live there. Bring a laptop and a focus block.
Each challenge has a hidden test suite or a property the grader checks against your run. You either pass, fail with a hint, or fail with a partial credit note. The canonical solution unlocks once you pass.
Each challenge ships with two graduated hints. The first nudges your approach, the second narrows the scope. The third option is a full walkthrough, available once you have spent enough time on it that the walkthrough actually teaches something.
Still have questions? Reach out and we will walk through your specifics.
Get in touchBe there when the first courses open.
Drop your email and we will let you know the moment the first courses go live. No marketing list, no drip. One short note when there is something to learn.
Thanks. Check your inbox to confirm academy launch updates.
Where AI security gets practiced.
Audits, research, and training from the team building the field's working toolchain.
LEARN MORE