AI Strategy, Infrastructure Architecture & Security Research
Connecting AI strategy with secure infrastructure, agentic systems, governance and evidence-backed implementation.
Enterprise infrastructure and cybersecurity form the foundation of the advisory work. See the career background for the scope of previous roles.
Explore the work, its sources and its limits. Authored research and software are not independent endorsements.
Based in Cologne, Germany. Available for engagements in Vienna and across DACH.
Calixa is an interactive assistant for exploring Roble Mumin’s professional background, projects and research. Answers should refer to public evidence and acknowledge uncertainty. Consult the original sources for qualifications, publication status and project limitations.
Explore the work, its sources and its limits. Authored research and software are not independent endorsements.
Pick the closest route. Each lands on a concrete surface, not a sales page.
Open Calixa — the career terminal. Ask about fit, paste a job description, get a structured review.
Browse the Evidence Library — public documents, frameworks, and supporting material behind the claims here.
Read the About page — career history, expertise areas, and how the practice came together over twenty years.
Use the contact form below for a brief or scoping conversation. Engagements are contracted off-platform.
The advisory approach connects AI risk assessment, technical documentation, human oversight and operational controls. Applicable duties depend on the system, organizational role and jurisdiction; this is not a guarantee of compliance.
Explore the work, its sources and its limits. Authored research and software are not independent endorsements.
Region-specific engagement pages, in German and English. Each city has both a DE and an EN page linked via hreflang.
Selected work in AI infrastructure, inference stability and computational research.
An open protocol for finding and selecting AI providers by capability and policy rather than a fixed vendor. The work connects a published architecture with client, directory and management implementations, while preserving participant control over execution.
Research into how load-related numerical variation may affect the stability of language-model inference. ISDT proposes testable hypotheses and a Prompt Weather Index for investigation, distinguishing output stability from answer correctness.
A layered computational ontology for quantum cosmology, exploring how physical processes might be described through a structured computational architecture. The work sets out its proposed framework, mathematical discussion and research questions.
An experimental number-theory framework investigating whether local prime-gap statistics and global density feedback can reproduce selected properties of prime sequences. The publication connects its research claims with executable verification and the accompanying LULZprime implementation.
An automated AI radio station created by my friend Arno Selhorst. A creative use of AI that I’m happy to share here. Tune in below.
Loading this player connects your browser to the independent SARC-FM website. Its own privacy practices apply.
Send the context. I will map it against strategy, delivery risk, governance, and evidence.