AI-First Senior Software Engineer

Nion is an IT consultancy with offices in Stockholm, Malmö, Gothenburg, Skopje, Bitola and Sofia, where 200+ IT specialists work together to deliver tailored digital solutions. Our clients operate in industries such as telecom, fintech and insurance, and we help organisations simplify complex challenges and create real business value. We're looking for an experienced AI-First Senior Software Engineer for our Stockholm office. This is not a "coder" role. Our position is deliberate: the era ahead needs software engineers , not software developers . AI has hyper-accelerated the software cycle — but it cannot hold accountability. The engineer does. With AI handling most code generation, human attention becomes the scarce resource and moves upstream and downstream of the keyboard: to priorities, trade-off decisions, governance and final judgment, while agents handle execution in between. Your job is to do what AI cannot — set direction, make the judgement calls, and keep acceleration from turning into compounding risk across functionality, security, compliance, maintainability and knowledge continuity. The Factory's goal is accelerated, secure and sustainable software delivery, and you are the person who holds that line. What you'll do - System design. Own the big picture — how components fit together, how data flows, how failures propagate. Modularity, standards and flexibility, not just the unit of code in front of you. - Domain modelling. Understand the client's business problem deeply enough to design the right solution, not just a working one. Widening the problem context is where the real advantage is. - Architectural patterns. Know which patterns to apply, when and why. As AI produces design artefacts faster than organisations can evaluate trade-offs, your role shifts toward decision engineering — choosing and validating, not just producing. - Process design — the core responsibility. Design the repeatable process by which software gets built, tested, deployed and maintained. In practice this means designing and orchestrating agentic coding flows, including patterns such as orchestrator–worker (an orchestrator decides sub-tasks, initiates them and synthesises results) and evaluator–optimizer (one model generates, a second evaluates and feeds back, and the loop iterates). You decide where a deterministic workflow is right versus a more autonomous agent. - Quality judgment. Know what good looks like, recognise when AI output is wrong, and have the discipline to reject it. Quality control moves from line-by-line review toward designing tests, feedback loops and validation checkpoints. - Tooling — the right tool for the job. Fluency across product-development tools, AI-acceleration (code-generation) tools, infrastructure tools and quality-control tools. The best results come when tools have full access to the specific codebase and knowledge base — and well-structured, standardised codebases yield the largest gains. - Accountability. You hold accountability — not the AI. Put bluntly: vibe coding is not software engineering. Your profile - 8+ years of hands-on software engineering, with deep fundamentals in architecture, system design, modularity and standards (non-negotiable — AI amplifies skill, it doesn't substitute for it) - An early adopter of AI-first development, with proven, in-production experience implementing AI-first software development processes — not just experimenting with tooling - Demonstrable experience designing and orchestrating agentic / AI-assisted development workflows (orchestrator–worker, evaluator–optimizer, and workflow-vs-agent decisions) - Strong critical evaluation of AI output and the discipline to reject it - Test design, feedback-loop design and validation skills over line-by-line review - Security, compliance and maintainability awareness in AI-accelerated pipelines - Infrastructure / CI-CD fluency and the ability to work across the full SDLC - Domain-modelling ability — translating a business problem into the right technical solution - Prompt and context engineering as a working skill, not a novelty - Strong collaboration and communication, with deep client and product integration - Fluent in Swedish and English, written and spoken Tools you'll work with - AI-acceleration: Claude Code, agentic coding tools and LLM platforms (Anthropic Claude, OpenAI / Azure OpenAI) - Languages & engineering: .NET, Java or Python at senior depth, with strong Git and code-review practice - Workflow orchestration: agent frameworks and multi-step orchestration (orchestrator–worker, evaluator–optimizer patterns) - Infrastructure: CI/CD pipelines, Docker, Kubernetes - Quality control: automated testing, evaluation and validation frameworks for AI-assisted output Meritorious - Anthropic Claude certification or equivalent AI platform credentials - Experience standing up AI-first processes at team or organisation scale - LLM evaluation, guardrails and testing strategy experience - Domain experience in telecom, fintech or insurance - Prior consulting or client-facing delivery experience What this role is not It is not a prompt-driven "code typist," and it is not someone who ships unreviewed AI output. Used incorrectly, AI tools make a codebase less reliable, and automated processes compound that risk. Your judgement, ownership and process design are the product. We're looking for someone who is - Professional and committed to delivering solutions that meet real business needs - Accountable and self-driven — you take ownership of the big things and the small things - Communicative and a team player; resilient under the cognitive load of carrying large context (we build teams with fail-over and redundancy, because AI burn-out is real) - Flexible and open to some travel in connection with client projects With us you get - A collaborative, flat organisation where support is a given. Your consultant manager acts as a mentor to ensure your profess