Principal Kernel Engineer

Principal Kernel Engineer

We are seeking a Principal Kernel Engineer. Driving deep kernel performance and power optimization for flagship mobile platforms. You will own performance across core kernel subsystems — memory management, scheduling, IPC, and synchronization — analysing behaviour on real silicon down to the microarchitectural level and landing production-shippable optimizations that improve responsiveness, throughput, and battery life. This is a hands-on role for an engineer who is at home in allocator internals, scheduler hot paths, and lock-contention traces, working alongside CPU microarchitects, runtime engineers, and AI-driven optimization tooling.

Key Responsibilities:

• Optimize memory management internals: buddy allocator, slab/SLUB, page reclaim and LRU/MGLRU, memory compaction and defragmentation, transparent huge pages, page-fault and TLB-shootdown paths, copy-on-write behaviour
• Drive scheduler performance: run-queue and load-balancing behaviour, energy-aware scheduling, wakeup latency, task placement on heterogeneous (big.LITTLE/DynamIQ) topologies, cpufreq/cpuidle governor interaction
• Optimize synchronization primitives and lock-heavy paths: futex fast/slow paths, mutexes and rwsems, spinlocks/qspinlocks, RCU, seqlocks, per-CPU data, memory ordering and barriers on ARM64
• Analyse and improve IPC and syscall paths: context-switch cost, binder-style IPC, shared memory, vDSO, interrupt/softirq handling
• Identify bottlenecks on flagship mobile workloads (gaming, day-of-use, camera, AI inference) using PMU counters, ftrace/perf, eBPF, lockdep/lockstat, and power measurement with mA/mAh attribution
• Land production-quality kernel patches from hypothesis through benchmarking to ship, with measured latency and energy impact
• Collaborate with CPU microarchitects on HW-SW co-design: exploit microarchitectural features in the kernel and shape future hardware through kernel-driven requirements
• Contribute to AI-assisted kernel optimization pipelines (automated analyze ? patch ? benchmark ? validate loops)

Person Specification:
Required:

• 10+ years of OS kernel development in C (Linux or comparable), with patches shipped to production systems at scale
• Expert-level understanding of MM internals: physical/virtual memory management, buddy and slab allocation, reclaim, compaction, page tables, TLB management
• Deep scheduler knowledge: CFS/EEVDF internals, load balancing, preemption, real-time classes, energy-aware scheduling
• Mastery of kernel synchronization: locking primitives, RCU, lock-free techniques, the ARM64 memory model and barrier semantics
• Proven kernel performance-analysis skills: PMU counters, ftrace/perf, eBPF, lock contention and latency analysis
• Strong ARM64 architecture grounding: exception levels, cache hierarchy and maintenance, TLBs, memory ordering

Desired:

• Mobile power engineering: DVFS governors, cpuidle, per-workload energy attribution
• Upstream Linux kernel contributions (mm, sched, locking, or related subsystems)
• Familiarity with HarmonyOS or Android common kernel
• Exposure to CPU microarchitecture and cycle-accurate modelling (gem5)
• Experience with LLM/agentic tooling applied to systems software