Excloud Platform Engineer
Jun 2023 to Aug 2026 | Remote
Public cloud with 4,000+ accounts, 380 active VMs, and 140 running managed Postgres clusters at handover; built and operated the core platform alongside the engineering lead and handled production escalations directly.
- Designed and built the block storage engine on SPDK and NVMe-oF, with a reconciler that converges database intent against live bdev, lvol, and RAID state so a crash mid-provision recovers instead of orphaning volumes.
- Implemented the VFIO-user data path from SPDK-backed RAID devices into QEMU guests, with per-VM NQNs and sockets, idempotent attach and detach, and readiness checks before a guest sees the disk.
- Built the Go compute control plane over QEMU/KVM: VM lifecycle APIs, rescue boot, volume attach and reap, subnet operations, and reconciliation after failed transitions.
- Shipped managed Kubernetes end to end: control-plane and worker bootstrap, OIDC and JWKS, Cilium and CoreDNS, a CSI driver with attach, mount, resize, and failover, and the Karpenter cloud provider that autoscales nodes.
- Wrote the shared Go layer the services are built on: an HTTP framework that derives OpenAPI 3.1 from Go types by reflection, a typed error taxonomy, cross-service domain types, and OpenTelemetry middleware, imported across 20+ internal services.
- Traced a borrowed-header lifetime bug through the HTTP/2 and HPACK layer of the Kubernetes API transport and fixed it without losing OpenTelemetry trace propagation.
- Built authoritative DNS with AXFR, TSIG, and weighted records, and an AF_PACKET ARP and NDP proxy with BPF filtering, VLAN handling, rate limits, GARP, and IPv6 DAD.
- Built the API-to-client pipeline from OpenAPI 3.1 across eight services, feeding the exc Go SDK and CLI plus a generated Terraform provider with 30 resources and 13 data sources.