Scrum Master

Scrum Master, Agentic Software Development

West Flanders, Belgium · Hybrid (4 days in the office) · Permanent

Most Scrum Masters spend their careers running a process someone else designed. In this role, you design it.

Our client is an independent software company that has run the operations of manufacturers in the label and packaging industry for over 30 years. Its management information system handles estimating, production planning, scheduling, shop-floor data and invoicing for customers in more than 15 countries. 

The business has around 100 people and is growing. It is now building a new cloud-native SaaS platform from the ground up, using AI-first principles. AI agents already plan, write and test real code alongside the developers. 

The opportunity:

You will lead agile delivery across the business and decide what Scrum looks like when part of the team is AI agents. This is not a role for running a standard playbook. The playbook does not exist yet, and writing it is the job.

Agentic experience is the deciding factor:

Hands-on experience of agentic software development matters more than anything else in this role. Strong Scrum credentials will not carry an application without it.

By agentic, we mean AI agents that plan, write, test and refactor code with real autonomy, such as Claude Code. We do not mean autocomplete tools or using a chatbot to answer questions.

By hands-on, we mean you have done this inside a real software team, not only read about it, coached around it or tried it on a side project. 

The strongest applicants will be able to show:

• Time spent working directly with coding agents, ideally Claude Code, on a codebase that ships to customers.
• A team they have helped move to agentic workflows, and what changed in planning, review, testing and definition of done as a result.
• Specific failures they have seen, such as agents changing the wrong files, over-broad permissions or code that passes tests but is wrong, and the guardrails they put in place afterwards.
• How they handled the new bottlenecks: specification quality, human review capacity and too much work in progress.
• A clear view on what delivery metrics still mean once story points stop reflecting effort.
• When you apply, give one or two concrete examples covering the points above. The more specific you are about what you personally did, the better.

What you'll do:

• Run sprint planning, stand-ups, reviews and retrospectives across two development teams in Belgium and Sweden, working in short sprints.
• Redesign those ceremonies for agentic work. For example, decide what a daily stand-up is for when agents work asynchronously and continuously.
• Rethink estimation, velocity and definition of done for work that agents generate and refactor at a different pace, with different failure modes.
• Coach developers as their role shifts from writing code to specifying intent, reviewing agent output and supervising agentic pipelines.
• Remove impediments, including agent-specific ones such as access and permissions, infrastructure bottlenecks and weak specifications.
• Work with Product Owners so backlogs and acceptance criteria are clear and testable enough for both people and agents to build against.
• Track and report delivery metrics that reflect real progress, replacing or supplementing story points where they no longer work.
• Build trust and shared practice between the two teams, including a common approach to how much autonomy agents are given.
• The tech stack is .NET, React and Entity Framework on Azure, with SQL Server and Elasticsearch. 
• The architecture is a modular monolith built on domain-driven design, vertical slices, multitenancy and API-first principles, with CI/CD and containers.

What you'll bring (the essentials)

• Hands-on agentic software development experience in a real team, as described above.
• Proven experience as a Scrum Master, Agile Coach, Delivery Lead or Tech Lead running agile delivery in a software product company, ideally B2B SaaS.
• Experience as a software developer is a strong advantage, as you will need to understand code review, testing and architecture well enough to challenge agent output.