System Development Engineer

The common nature of the work:

Every one is a System Development Engineer (Contractor) position, Seattle-based, L5–L6 equivalent, 1 opening each, sitting on the Region Flexibility Migration (RFM) team inside the client's Intelligent Cloud Hosting (ICON) organization.

They all serve a single 2026 initiative: DRAM (memory) footprint reduction across the client's production service fleet — essentially cost/efficiency and performance engineering, not new-feature or product development.

The job is to find where the client's services and clusters are wasting RAM, optimize it away without breaking anything, validate the savings, and turn each win into a reusable playbook.

What differs is the lever — the layer of the stack each role targets:

• EKS Cluster Management — the container/orchestration layer. Manage Kubernetes (EKS) clusters end-to-end and profile containerized workloads: right-size pod resource requests/limits, tune VPA and node pools, and hunt memory waste at the cgroup/runtime level.
• EMR Cluster Operations & Spark Tuning — the infrastructure/config layer for big-data. Right-size EMR clusters and instance families (e.g., memory-optimized R-type compute-optimized C-type), and tune Spark/YARN settings at the cluster level.
• Memory Optimization (Spark) — the application/code layer for big-data. Deep-dive and refactor the actual PySpark / Spark / Hive / Pig job code — rewriting inefficient transformations, fixing partition skew, tuning caching and shuffles.
• JDK Migration & Heap Right-Sizing — the language-runtime layer. Migrate Java services across JDK versions (17 21, 21 25), resolve migration blockers, and right-size JVM heaps / exploit new algorithms (Zgenerational, etc.).
• Rustification & Service Migration — the rewrite layer. Rebuild memory-bound Java services in Rust to eliminate JVM overhead entirely (targeting ~50%+ memory reduction), proving functional equivalence via oracle/differential testing and doing a safe canary cutover.