Over a decade at IBM, I served as a Senior Software Engineer on PowerVC — IBM's enterprise cloud management platform built on OpenStack, designed to orchestrate virtualised workloads on IBM Power Systems hardware for Fortune 500 data centres. The platform manages compute (VMs), block storage (volumes), networking, and image lifecycle as a unified API surface, sitting above IBM's POWER hypervisor stack.
My primary domain was storage driver engineering — the integration layer between the cloud storage orchestration service and enterprise SAN arrays. I wrote and maintained the Python drivers that translate abstract cloud API calls (create a volume, attach it to a server, live-migrate, snapshot, replicate) into vendor-specific storage operations for: IBM Storwize/SVC, IBM DS8000, IBM FlashSystem, EMC VMAX/VMAX3, EMC VNX, Hitachi VSP, IBM XIV, and IBM Spectrum Scale (GPFS). This required deep knowledge of Fibre Channel and iSCSI connectivity, LUN masking, HBA zoning, and multi-path I/O in HA environments.
A major parallel workstream was virtual machine lifecycle management: owning the control-plane code for deploying, capturing (templating), migrating, and deleting VMs on IBM Power hardware via the Hardware Management Console (HMC). This included handling boot-disk attachment sequences across two connectivity models — NPIV (virtual Fibre Channel) and vSCSI (virtual SCSI over a shared I/O server) — as well as live and cold migration of VMs across physical hosts while keeping storage paths intact.
I also owned SAN fabric automation: automatically managing Fibre Channel switch zones (Brocade and Cisco) when VMs are provisioned or migrated, so that storage paths activate without manual switch configuration. Related work covered port-group policy enforcement and fabric consistency validation across large enterprise deployments.
A third domain was cloud-init and OS image lifecycle. Cloud-init is the industry-standard mechanism by which a newly-deployed virtual machine self-configures (hostname, network interfaces, SSH keys, user data) on first boot. I ported and qualified this toolchain across RHEL, SLES, AIX, and Ubuntu guests on IBM Power hardware, and owned the image capture / export / import pipeline.
From 2022 onward my scope expanded significantly to feature engineering and architecture. Shipped features include: bulk volume delete (scheduler-aware parallel deletion pipeline), attach/detach performance optimisation via DB index tuning and batched API calls, image appliance node export/import workflows, VM import REST APIs, consistency group operations, initiator port-group auto-management, and backup-node upgrade flows. I drove these end-to-end: design, implementation, code review, unit & functional tests, performance benchmarking, and post-GA customer escalation support.
The 1,650 defects personally investigated reflect a broader role beyond ownership — acting as the primary escalation point for field-reported production issues from enterprise clients worldwide. Each of those investigations involved reading diagnostic logs, reproducing failures in lab environments, providing root-cause analysis, and either authoring the fix or guiding the owning engineer to resolution.
Across 845 resolved bugs, 145 feature stories, and 59 engineering tasks the work spans storage drivers, hypervisor management, SAN networking, cloud-init, UI/API layers, and distributed-system reliability across the PowerVC platform.
| Year | Bugs Owned | Stories | Tasks | Owned Total | Bugs Investigated | Grand Total |
|---|
Storage drivers & backend integration (~35% of bug volume) — Python drivers that bridge cloud storage APIs to physical SAN arrays. Operations include volume CRUD, attach/detach, live migration reattachment, snapshot/clone, and replication across IBM SVC, DS8000, FlashSystem, EMC VMAX, EMC VNX, Hitachi VSP, XIV, and IBM Spectrum Scale (GPFS).
VM lifecycle & hypervisor management (~25% of bug volume) — Control-plane logic for deploying, capturing, migrating, and deleting virtual machines on IBM Power hardware via HMC. Covers boot-disk sequencing for both NPIV (virtual Fibre Channel) and vSCSI (shared I/O server) connectivity models, multi-disk live migrations, and VIOS management.
SAN fabric & zoning automation (~15% of bug volume) — Automatic management of Fibre Channel switch zones (Brocade, Cisco) during VM provisioning and migration. Covers port-group policy enforcement, fabric consistency validation, and initiator port-group lifecycle management.
Cloud-init & OS image lifecycle (~12% of bug volume) — Porting and qualifying cloud-init (first-boot VM self-configuration) across RHEL, SLES, AIX, and Ubuntu on IBM Power. Also covers image capture/export/import tooling and image currency across OS releases.
Feature engineering & architecture (~55% of story volume) — Larger-scope delivered features: bulk volume delete pipeline, attach/detach performance via DB optimisation, image appliance node workflows, VM import APIs, consistency groups, backup-node upgrade flows, and UI/GraphQL contributions.
Observability, testing & platform reliability (remainder) — Cross-cutting work: health-check APIs, integration-test framework contributions, quota validation, and root-cause analysis support for enterprise customer escalations across storage, compute, and networking.