What I work on

• 🤖  AI Infrastructure: LLM serving, GPU scheduling & sharing, multi-node inference
• 🧱  Hyper-Converged Platforms: compute, storage, and VMs on one Kubernetes stack
• 🛡️  High Availability: failover orchestration that survives infrastructure failures
• ⚡  High-Performance Networking: DPDK, eBPF, zero-copy I/O
• 🔬  Research: published work on 5G core networks & high-performance I/O (IEEE NetSoft & CloudNet, 2023)
• 📫  Reach me at jerry10127@gmail.com

Software Engineer · Phison Electronics · Hyper-Converged Infrastructure (HCI) · Sep 2026 – Present

• 🧱  Building Phison's HCI platform, an on-prem Kubernetes stack (RKE2, Cilium, kube-vip, envoy-gateway) unifying compute, storage, and AI inference
• 🤖  Working on the AI inference layer: KServe, llm-d & vLLM, multi-node inference with leader-worker-set, and GPU sharing with HAMi
• 💾  Integrating Rook Ceph storage and aiDAPTIV+ KV-cache tiering on Phison ai100 NVMe
• 🏢  Supporting a multi-tenant platform with Keycloak SSO, KubeVirt VMs, and managed external clusters

Software Engineer · Saviah Technologies Inc. · Aug 2023 – Aug 2026

• 🚀  Architected a failover orchestration Kubernetes operator — cut service recovery 7.2× (180s → 25s) at 99.9% availability for 50k+ concurrent sessions
• ⚡  Built an eBPF-based load-balancing dataplane — reduced worst-case rerouting latency 90% (10s → 1s) during control-plane failures
• 🌐  Designed a topology management system (RESTful APIs) serving 100k+ subscribers across 10+ edge nodes
• 🔧  Automated the full infra lifecycle with an Ansible framework standardizing 100+ production environments
• 🧪  Built a CI/CD stress-testing platform simulating 10k+ sessions, used by a 20+ engineer team to catch SLO regressions

💬 Let's build resilient systems together!