Principal Software Engineer (Kubernetes Control Plane & AKS Flex)

Quick Facts

• Role: Principal Software Engineer (AKS Flex / Kubernetes control plane scalability)

• Focus: Scaling etcd and the Kubernetes control plane; building platform capabilities for flex/remote nodes via provisioning, lifecycle automation, and Kubernetes-native APIs

Description

You will lead the architecture and scaling of etcd and the Kubernetes control plane so a single cluster can manage far larger, more distributed, and heterogeneous worker node fleets. You will also build the platform behind AKS Flex and Unbounded, enabling secure joining, provisioning, lifecycle management, and declarative remote-node control with Kubernetes-native APIs. The work targets distributed-systems correctness, durability, performance, security, and production reliability.

Responsibilities

• Own and evolve the architecture for control plane scalability, flex nodes, and the Unbounded platform; help shape a multi-year technical roadmap

• Lead cross-organization designs across AKS, Azure compute, storage, networking, and identity

• Make and defend architectural decisions for correctness, durability, security, scale limits, and failure modes; set design review and quality bar

• Represent upstream open-source communities (e.g., Kubernetes SIG Scalability, SIG etcd, SIG API Machinery) and the Unbounded project

• Mentor senior engineers and raise overall technical capability

• Improve etcd and Kubernetes API server performance, reliability, and efficiency under high scale, object churn, and watch traffic

• Optimize control plane bottlenecks (write amplification, compaction/defragmentation, watch fan-out, list/relist storms, request prioritization, storage/memory pressure)

• Design and validate control plane storage layer changes (partitioning, offloading, caching) while preserving correctness/durability guarantees

• Build scale/performance testing infrastructure, benchmarks, and SLOs to measure control plane limits and prevent regressions

• Automate safe etcd operations at fleet scale (backup/restore, member replacement, quorum recovery, upgrades, repair)

• Design and build secure remote node joining so nodes behave like standard node pools across regions, on-prem, edge, and other clouds

• Implement provisioning and day-2 lifecycle automation across substrates (SSH bootstrap, cloud API provisioning, bare-metal PXE with BMC power management and TPM-based attestation)

• Create Kubernetes-native APIs, controllers, and CRDs for declarative remote machine management at fleet scale

• Handle remote-node control plane challenges (latency, intermittent connectivity, heartbeats/status churn, node identity/authentication, secure bootstrap, safe upgrades across heterogeneous hardware)

• Contribute to and help lead the Unbounded open-source project (design, code reviews, community engagement)

• Participate in production operations/on-call and turn operational learnings into durable improvements

• Partner with AKS and Azure teams to deliver secure, compliant, resilient infrastructure

• Drive engineering excellence through technical design, code reviews, and reliability/operational standards

Benefits

• Opportunity to define early architecture for a new product area and influence hyperscale managed Kubernetes “without boundaries”

• Chance to represent Microsoft in upstream Kubernetes and Unbounded communities

• Work directly enabling capacity for GPU-heavy AI workloads and data-residency-constrained customers