I transitioned into DevOps from an Arts background and have about 3.3 years of experience across four companies. My work has included AWS, Kubernetes, Terraform, CI/CD, automation, and I have earned AWS and CKA certifications. I recently joined a large organization, but the work culture is exhausting and I am not learning as much as I expected. It has made me question whether I should continue on the same path or make a more deliberate career change. Over the next three to five years, should I focus on becoming a stronger Platform Engineer or SRE, move into AI/LLM or MLOps, pursue Solutions Architecture, specialize in security, target international remote roles, or prepare to relocate abroad? If you were starting over with similar experience, what roadmap would you follow and why?
5 Answers
Solutions Architecture could be a strong long-term option if you enjoy understanding customer problems, explaining trade-offs, and translating requirements into workable designs. Several years of DevOps and distributed-systems experience can provide a solid technical foundation, but you will also need communication, stakeholder management, system design, and business skills. The job may involve more meetings and coordination and less hands-on implementation, so make sure that matches the kind of work you want.
Before choosing a trendy title, figure out what kind of work actually gives you energy. Three years is still early in DevOps, and the field is broad enough that you may simply not have gone deep in one area yet. Try to distinguish between disliking DevOps itself and disliking your current company. Talk to teams working on networking, security, observability, or internal platforms and volunteer for projects with them. Then choose one direction you genuinely enjoy and build depth instead of constantly chasing the next role.
Your experience already points naturally toward Platform Engineering or SRE. Focus on operating real systems, networking, reliability, observability, incident response, and platform design rather than only assembling YAML and pipelines. AI will change how infrastructure work is done, but it will not remove the need for people who understand production systems, trade-offs, maintenance, and business requirements. Security and MLOps are good adjacent paths if they genuinely interest you, but do not try to pursue all of them at once.
Do not assume that switching companies is automatically the problem, but be more selective about the next move. Study the kinds of roles you want at companies you respect, then use their requirements to guide your learning. In interviews, ask about the team’s actual responsibilities, on-call load, engineering standards, learning opportunities, and how much time is spent on repetitive support work. A better environment with room to experiment may restore your interest without requiring a complete career change.
AI infrastructure and MLOps are worth exploring if you like the technical side of AI systems. Your cloud, Kubernetes, automation, and operations background transfers well to deploying models, managing inference workloads, monitoring AI services, and building reliable data or model pipelines. I would treat AI as an extension of your existing systems knowledge rather than abandoning everything to become a machine-learning researcher. Build a small end-to-end project first and see whether you enjoy the work before committing your whole roadmap.
A useful next step is to pick one axis—platform, reliability, or security—and spend the next couple of years owning meaningful systems in that area. Breadth is valuable, but depth is what usually removes the feeling of being lost.