From monthly releases everyone dreaded, to daily
4 frontend developers · 10–15 in the team
- Daily releases instead of monthly
- Agentic coding across the whole team
- Micro-frontends and GraphQL in front of legacy
The team builds one of Grundfos' digital customer platforms: an IoT platform that monitors and controls physical equipment. A missing alarm or a wrong command can cost millions in property damage and, at worst, personal injury, so users must be able to trust the data even when the connection is unstable. Technically it is React frontends on top of a microservice-based backend with GraphQL as the unifying layer. I joined as frontend tech lead and over time took responsibility for process, releases and technical enablement across the team.
“You have played a central role in getting the product to a good place. You came in and got refinement, deployments and GraphQL under control, and you have helped us work more deliberately with our technical enablers. Whenever we have had hard bugs or performance problems, you have presented several approaches and then recommended one. I have used that several times as an example to others of what good behaviour looks like. Your judgement is highly trusted in the team, and for good reason. But what I appreciate most is that you keep pushing forward, most recently with AI. That drive is rare, and it is contagious.”
Releases nobody is nervous about any more
When I arrived, releases happened monthly, the team dreaded them, and each one was followed by a wave of bugs. The backend was built as microservices but behaved like a distributed monolith, with lockstep deployments and no real way to roll back.
Today we release daily across all services. Many people and many decisions made that happen, but the enablers I introduced were:
- Swagger specs and enforced OASDiff checks on every backend
- API and GraphQL schema registries to manage breaking changes and see schema usage across services and environments
- Error monitoring for proactive bug handling
- A release process with easy rollbacks and cherry-picking
- Feature flags for granular rollouts and instant rollbacks
GraphQL used properly
The team used GraphQL as a REST proxy. Over time I trained the team in the right way to use it and had the codebase reworked to play to GraphQL's strengths: one schema, precise queries and type safety all the way to the client.
Agentic coding across the whole team
In 2026 I was the primary driver of introducing agentic coding in the team, including the guardrails, review discipline and tooling that come with it. It has spread to teams around us and given our non-developers new options.
One example: I built a setup that let our product owner build mockups of features directly in the React codebase and catch UX problems before anything was built. Handovers got sharper, and calendar time from idea to release dropped by up to 80%.
Teams in and out, without the quality wobbling
Along the way we onboarded and offboarded an entire external team of frontend and backend developers. I owned the onboarding of the frontend developers and the guardrails that kept users from noticing any dip in quality.
Later we introduced a micro-frontend architecture so sister teams at Grundfos can contribute with NPM packages or whole micro-frontends rendered in the main app. Process, tooling and setup were my responsibility.







