About
I'm Amine.
I help teams build LLM systems
that don't become security incidents.
My background is in securing critical systems across regulated industries — infrastructure where failure has real consequences. I've worked across several sectors where security isn't optional and the cost of getting it wrong is high.
More recently, I've been working with enterprise clients on building and deploying LLM-based systems — model selection, fine-tuning, RAG pipelines, agentic workflows, production infrastructure on Kubernetes. That's where the two worlds collide.
That dual vantage point matters because LLM vulnerabilities don't look like traditional security flaws. Prompt injection, data leakage through model outputs, privilege escalation in agentic systems — these emerge from how the technology behaves, not just how it's configured.
The gap I kept running into: engineering teams building sophisticated AI systems without security expertise in the room, and security teams asked to review systems they didn't fully understand. Neither group was wrong — the knowledge to bridge that gap just didn't exist yet.
That's what I built Molntek to do. I work with organizations that need someone who can speak both languages fluently — deep enough in AI engineering to understand how these systems actually behave, and security-focused enough to find where they break.
What makes this different
I've built these systems, not just reviewed them
Prompt injection, data extraction paths, and agentic risk aren't theoretical to me. I understand them because I've built the systems where these vulnerabilities live — not just from reading OWASP documentation.
Findings you can act on
My output is written for engineers: specific vulnerabilities, how to reproduce them, and concrete remediation steps. Not a compliance checklist, not boilerplate recommendations.
Narrow scope, deeper expertise
I work on AI and LLM security specifically — not general cybersecurity, not compliance box-checking. A narrower focus means I'm not spreading expertise thin across every category of risk.
How I work
You talk to me directly. No account managers, no junior staff doing the work while a senior signs off. If you engage with Molntek, you engage with me.
I'll tell you if it's not the right fit. A bad engagement helps no one. If what you need isn't what I do, I'll say so on the first call.
The goal is a safer system, not a finished report. I care whether your team understands the risks and knows what to do about them — not whether the PDF is thorough enough to file away.
Want to know if I can help?
30 minutes. I'll ask about your system and tell you honestly what I think.