Join UMMRO
Help us build the platform that makes AI systems safe, robust, and compliant. We are a small team working on hard problems at the intersection of AI safety and regulation.
Our Story
UMMRO began as doctoral research into compliance engineering — a systematic way to prove that AI systems meet the robustness requirements regulators now demand.
That research became a platform: an orchestrator that tests 25+ language models with automated reframing strategies, soft-refusal detection, and safety scoring, mapped directly to EU AI Act Article 15.
Today we are turning that foundation into the compliance layer for the AI era, serving compliance teams and AI-safety researchers around the world.
Key Milestones
- 2024UMMRO research project founded; core reframing methodology developed
- 2025Platform launched with 25+ model integrations and 450+ test exercises
- 2026EU AI Act Article 15 assessment suite and CART framework released
- 2026+Scaling to enterprise compliance teams across Europe and beyond
Our Values
The principles that guide how we build and how we work together.
Security First
Every design decision is a security decision. Protecting customer data and testing responsibly are non-negotiable.
Intellectual Rigor
Our product is grounded in peer-reviewed research, and we hold our methods to academic standards of evidence.
User Empathy
Compliance is stressful. We build tools that make hard regulatory work clear, calm, and manageable.
Innovation
AI threats evolve weekly. We stay ahead by experimenting boldly and shipping what works.
Collaboration
We win as a team. Open discussion, honest feedback, and shared ownership — over silos and egos.
Execution Excellence
Ideas matter only when shipped. We move fast, finish what we start, and sweat the details.
Why Join UMMRO
Impact at Scale
Your work directly shapes how organizations across Europe and beyond deploy AI safely and legally.
Technical Excellence
Work on a research-grade orchestration platform spanning 25+ LLMs, adversarial testing, and safety scoring.
Constant Learning
Operate at the frontier of AI safety, red-teaming, and regulation — with the time and budget to keep learning.
Company Building
Join early enough to shape the product, the culture, and the company itself.
Hard Problems
Adversarial robustness, refusal detection, multi-model orchestration — problems without textbook answers.
Competitive Compensation
Meaningful equity, competitive salaries, and flexible remote-first working.