Services
Infrastructure, Governance, and Capacity Building
I work at the intersection of public administration and software engineering. The focus is on building AI infrastructure that is technically independent, understands Icelandic context, and meets requirements for privacy and security.
At the University of Akureyri, I led the development of Iceland’s first comprehensive university AI policy and managed production systems such as BORG and Arctic Tracker. That work is grounded in the conviction that policy without technical capacity to enforce it is meaningless.
Below is an overview of the methodology and services I offer to institutions and businesses that want to build on their own terms.
Methodology & Architecture
The systems I build follow four design principles.
Vendor Agnostic
Infrastructure is designed so that language models can be swapped without rebuilding the system. This reduces risk if vendors change terms or pricing.
Icelandic Language & Cultural Literacy
Systems use custom prompts and RAG (Retrieval Augmented Generation) to work in Icelandic with the nuances that matter in governance.
Data Sovereignty
Sensitive data and decision-making processes are on infrastructure the organisation controls.
Transparency by Design
Every system is documented so that technical teams can trace how a result was reached. A prerequisite for accountability over system output.
The Proof is in the Architecture
Sumarhús Alpha v2.4
I do not sell theories. I sell the systems I live in. This is the architecture of my own infrastructure, managing everything from thesis research to home heating. Self-hosted, sovereign, documented to the line of code.
Cloud APIs
Anthropic
Claude Opus 4.6
Gemini
OpenRouter
Multi-model
MacBook Pro
PAI Command Center
Infrastructure
Hercules
GPU Compute
Proxmox
Hypervisor
Docker VM
Container Host
PocketBase
Database + API
LAN · IoT
S7-1200 PLC
Hot Tub Control
Home Assistant
Smart Home Bridge
When I build for you, I bring this level of rigor.
Handover
All projects end with handover to the organisation. The goal is for the internal team to run the system independently.
The handover package includes:
- Architecture Documentation — A technical overview enabling in-house developers to take over maintenance and development.
- Controls — System prompts that govern model behaviour, delivered as plain text.
- Operations Manual — Instructions for updates, cost management, and quality control.
Public Institutions
Assessment & Policy Development
Mapping the current state of the organisation with respect to AI. Accountability structures, dependent processes, and legal compliance.
- Mapping current usage and shadow AI
- Risk assessment based on upcoming EU AI Act
- Development of actionable AI policy
Ministries, universities, and public institutions.
Governance Frameworks
The EU AI Act requires institutions to have control over their systems. I assist in establishing governance processes.
- Implementation of the BORG governance framework
- Preparation for high-risk AI system requirements
- Design of oversight processes for automated decision-making
Leaders responsible for regulatory compliance.
Businesses
AI infrastructure for businesses that want to own their systems and protect intellectual property.
- Local AI models to protect sensitive data
- Workflow automation (Agentic Workflows)
- API integrations independent of individual vendors
Training & Education
Workshops & Courses
Practical courses for executives and specialists. How the technology works, where it fails, and how to govern it.
In Icelandic or English.
Team Coaching
Working with key staff to set up workflows and develop in-house solutions.
Technical teams and middle management.
Speaking
Talks on AI, its ethics, and societal impact.
Example topics:
Responsibility Fog
How automation distributes accountability so that nobody seems to bear it.
Cognitive Debt
What happens when we outsource judgment to systems we don’t understand.
The Investment Gap
The mismatch between what investors want (AGI) and what businesses need (reliable automation).
Technical Sovereignty
Why we need to own our infrastructure in a digital world.
Talks are tailored to each event.
Open Knowledge
Foundational knowledge about AI should not be hidden behind paywalls. I consider it a researcher’s duty to share knowledge about the impact and workings of this technology with the public.
That’s why I host a podcast, write articles, and publish educational material openly. What I sell is specialist work building complex systems, not access to general information.
Those who understand the technology are better equipped to make informed decisions. That is a prerequisite for democracy in a technology-driven society.
Background
I work full-time and am involved in several projects simultaneously. This is the background I bring.
- Tutor and speaker at Oxford Lifelong Learning — AI Engineering and Low-Code Data Science
- Co-author, The Irreducible Human: Life, Value, and Meaning in the Age of AI (Brill Publishers)
- MA thesis: Beyond Fragmentation: A Life-Value Alternative for AI Governance
- Expert Panel Member, Rannís (Icelandic Centre for Research)
- Author of Iceland's first comprehensive university AI policy
- Erdos Research Fellowship (ERF.ai)
- Builder of BORG (AI governance platform), Arctic Tracker (conservation analytics, 473,000+ CITES records), and Gjöll (fire safety systems)
- Speaker at Equality Days 2026
- Host, Temjum tæknina (Taming Technology) podcast
- Previous career: 16 years in emergency services
If you are working through AI-related challenges — whether in policy, governance, or infrastructure — and would like to discuss them, I welcome the conversation.