AI governance
Clarify decision authority, oversight responsibilities, and evidence requirements for AI adoption and use.
AI governanceSecurity architectureAccountable decisions
Consulting to help organizations assess AI risk, strengthen security architecture, and establish clear governance.
Discuss a consulting engagementOur Mission
To help organizations use AI responsibly while protecting people, preserving human authority, and making consequential decisions accountable. Through consulting, research, and education, Odins LLC turns governance and security principles into practical controls.
Our Values
We put people, privacy, and security at the center of our work. We ground recommendations in evidence, make limitations clear, and connect decisions to accountable ownership. We challenge assumptions, document what we learn, and improve when evidence reveals a gap. As technology gains capability, human authority and responsibility must remain explicit.
Consulting
Security architecture
Review system boundaries, access controls, and failure paths to identify security gaps and practical improvements.
Risk advisory
Assess AI risks, prioritize treatments, and document residual risk and accountable ownership.
Research & Projects
Explore how Odins LLC approaches AI defense, accountable governance, and controlled autonomy.
AEGIS
Defense that checks itself
Designed to bring coordinated defense layers, threat detection, and continuous validation of its own defenses together. The focus includes both incoming threats and the reliability of the defenses themselves.
Explore AEGIS on GitHubOQGF
Make accountability part of the architecture
Odins Quantitative Governance Framework connects AI risk decisions to clear authority and auditable evidence, helping answer who can act, under what conditions, and who owns the risk.
Explore OQGF on GitHubBROKKR
AI can propose an action. Permission is a separate decision.
A governed autonomous coding agent under active development. Deterministic authorization gates sit between model proposals and execution, making control over actions central to the architecture.
Explore BROKKR on GitHubAI Governance Foundations
Understand the decisions behind responsible AI
Explore practical teaching modules covering AI governance, risk, compliance, security, and cryptography. Build the foundations needed to ask better questions and make informed decisions about AI.
Explore the teaching modulesAbout Jeremy
Jeremy Rose
Founder of Odins LLC
I'm a security and systems professional with a background spanning COMSEC, cryptographic key management, federal systems, public safety technology, operational risk, and now AI security and governance.
My work today focuses heavily on the problems emerging as AI systems become more capable and autonomous: prompt injection, adversarial behavior, capability and authority boundaries, containment, recursive risk, threat modeling, and how organizations maintain control when AI can reason and act at machine speed.
I'm also the founder of Odins LLC and the creator of Odins Quantitative Governance Framework (OQGF), where I research and develop approaches for governing AI systems without placing the model itself in the trust path. A principle that guides much of my work is simple: the model may propose, but deterministic controls should govern consequential actions.
My working style is grounded in evidence. I document constantly as part of the process. Notes give me continuity across complex work, help make results reproducible, and give me a record to return to when something fails. When a problem appears, I want to be able to trace what changed, identify the assumption or gap that produced it, correct it, and turn what I learned into a stronger process or regression test.
That same mindset shapes how I approach AI security: understand the system, challenge its assumptions, find where the boundaries fail, preserve the evidence, and make the next iteration harder to break.
I'm particularly interested in AI security, AI red teaming, adversarial ML, autonomous agent security, AI governance, model risk, and the engineering controls required to deploy increasingly capable AI responsibly.
Let's talk
Have a question about AI governance, security architecture, or risk? Get in touch to discuss consulting, research collaboration, or speaking opportunities.
jeremy@odinheim.ioWasilla, Alaska