SoftwareDevelopmentEngineerII-AmazonMSK,ManagedStreamingKafka(MSK),MSKInfrastructureManagement

Come keep one of the world's largest Apache Kafka fleets healthy, secure, and always on. On Amazon MSK, you will build the automation that maintains hundreds of thousands of streaming hosts, so customers never have to think about the infrastructure underneath their applications. This is infrastructure engineering at a scale where every change has to be safe by design. MSK is stateful: every host holds customer data that must stay replicated and in sync, so routine maintenance is never as simple as rebooting or replacing a host. Patching, repairing, and replacing nodes across a fleet this size means coordinating each action against the availability guarantees customers depend on, so that maintenance stays invisible to them. You will design and build the automation that does this, turning hard operational problems into systems that run themselves. If you enjoy distributed systems, large-scale automation, and work whose reliability countless streaming applications quietly depend on, this is a rare place to do it. Key job responsibilities - Design, build, and operate automation that patches and maintains hundreds of thousands of stateful hosts, keeping fleet maintenance invisible to customers. - Build systems that automatically detect unhealthy hosts and remediate them, balancing fast recovery against avoiding needless disruption. - Develop rollout and rollback mechanisms that keep the blast radius of any change small at fleet scale, and that let changes be tested before they reach customers and reversed if something goes wrong. - Own your services end to end: take part in on-call, debug production issues, and continually reduce the manual effort needed to operate the fleet. - Write design documents, collaborate with engineers across MSK, and raise the engineering bar through design and code reviews. A day in the life No two days look the same, but most blend building with operating. You might spend the morning designing a safer way to roll out a fleet-wide change, pair