Lavenix is a hands-on, engineering-first security studio from Amsterdam. We help teams ship safer LLM apps and connected devices by turning risk into working tests, small code fixes, and audit-ready evidence.
What we do
We design and secure LLM-driven systems and connected products end-to-end: threat-model and harden AI pipelines (prompt safety, I/O sanitisation, context isolation, PII handling, evals and red-teaming), probe real attack surfaces (AI, OT/(I)IoT/robotics, web apps, APIs), and review system architectures and code for resilience, reliability, and cost. We advise on stack and protocol choices for secure AI integration, apply practical cryptography (PKI, TPM/HSM, key management), align builds to relevant standards (ETSI/RED/CRA/NIS2 and emerging AI governance), establish secure SDLC practices, set up vulnerability/incident handling that actually runs, and run targeted assessments during design and implementation to catch defects early.
How we work
Fixed-fee sprints. We deliver scripts, tests, and code; your team (or DevOps partner) wires them into CI/CD. Heavy governance is partnered out.
Experience
Offensive security + system architecture across AI/LLM, IoT/OT/EV, complex APIs, applied crypto, and provisioning/manufacturing. The output is always the same: tests that run, code that hardens, evidence that stands up.