Full time Junior to mid-level On site
AI Engineer
You steer the AI. You do not compete with it.
Full time, based in Virum. You set the architecture, specify the task and review what comes back — the agent writes the code.
All coding at umage.ai runs through AI agents. Not as autocomplete — as the primary way we build software.
About the role
You will not be measured on how much code you type by hand, but on whether you understand the system well enough to steer an agent through it safely — from the first spec to code in production.
In practice that means making the decisions that come before the code: architecture, data model, scope. And reviewing what comes back with the same scepticism you would apply to a colleague's work.
About you
You are junior to mid-level in experience but solidly grounded: you understand coding, architecture, data and algorithms, even when you are not writing all the code yourself. You are curious, self-directed, and genuinely — not dutifully — interested in AI and LLMs.
About us
umage.ai is a newly founded software and AI house started by three experienced partners. We work across three areas: agentic development (AI agents as a product and as part of client solutions), AI in business (AI transformation at our clients), and AI sovereignty (EU-based, self-hosted independence from US big tech — in what we deliver, and in how we run the company itself).
People define architecture, quality, business understanding and direction; the agent writes the code. This is not an experiment for us, it is the working week. And in a small, close team the distance from idea to action is short: you become one of our first AI Engineers, and you help shape how the role develops.
What you will work on
-
Architecture and data models
For client work — the decisions that come before any code, and that an agent will not make for you.
-
Steering AI agents
Claude Code and similar, across the whole development flow: spec, implementation, tests, review.
-
Quality assurance of AI-generated code
Correctness, security, performance and maintainability. This is where the role stands or falls.
-
Data, algorithms and integrations
Where you need to understand the problem deeply enough to guide the agent correctly.
-
Working directly with clients
On agentic systems, AI in business, and AI sovereignty.
-
Our own internal tooling
Agent platform and internal automation — our own tooling is a product too.
What we expect
- Solid grasp of architecture, data modelling and algorithms — whatever the language or stack
- Active use of AI and LLMs in development (Claude Code, Copilot, Cursor or similar) — not just tried, but used to build something
- Able to read and understand code quickly, including code you did not write
- Comfortable without fixed frameworks: we are a small team still shaping our own processes
- You live in Denmark and are based at our Virum office
Nice to have
- Experience with agent frameworks, MCP or comparable agentic tooling
- Familiarity with cloud and infrastructure (Azure, AWS, or EU-based and self-hosted alternatives)
- Hands-on experience with Python, TypeScript or .NET
- An interest in AI sovereignty and European alternatives to US big tech
- Danish is an advantage, not a requirement — a few clients hold meetings in Danish
Applications
Apply for this role
Send your CV and a few lines on why this role in particular. If you have built something with an agent, send a link. If you have questions first, call.
We treat your application confidentially and share it only with the people involved in the hiring decision, which may include an external adviser. We keep applications for up to six months unless you ask us not to. Read the privacy policy
Questions before applying?
Contact Morten Aslo-Østergaard, Co-founder & CEO
Send application +45 7071 3333