Darko Fabijan

At Semaphore, we’ve spent over 15 years helping others optimize their CI/CD workflows, but tackling our own architecture’s growing pains was a whole different story. As our platform grew to over 30 microservices hosted in separate repositories, the cracks began to show. Onboarding was painful, dependency updates were tedious, and inconsistent pipelines made maintaining our own CI/CD workflows a chore.

This talk presents a journey from a fragmented and disparate microservices architecture to a unified and consolidated build system with monorepo.

I will discuss the reasons and challenges that led to this transition, as well as the benefits and problems encountered along the way. I'll also share future plans for continuous improvement while highlighting wins such as improved build speed, simplified local testing, and easier management of dependencies and the Software Delivery Lifecycle.

This talk isn’t just a playbook for standardizing a microservices architecture; it’s an honest tale from the trenches, filled with actionable insights and lessons learned for DevOps practitioners facing similar scaling challenges.

Key takeaways:

• Why we moved to a monorepo structure and how we unified our CI/CD pipelines, dependency management, and Docker image builds

• The unexpected challenges we faced—and the hard lessons learned—from inconsistent folder structures to managing cross-microservice dependencies.

• How these changes doubled our build speeds, simplified local testing, streamlined SDLC processes, and made on-prem releases dramatically easier.

• What’s next in our journey to keep improving and scaling effectively.