Fault-Tolerant Credential Rotation

Automating rotation of previously non-expiring credentials across 12+ infrastructure components, spanning Fleet Lifecycle and SDDC Lifecycle microservices in Kubernetes. Per-instance schedulers use lock-based coordination, pluggable rotation strategies, retries, and cleanup safeguards.

Designed and led end-to-end delivery

Scaling Infrastructure Management to 4,000 Hosts

Raising the supported capacity of VCF SDDC Manager by introducing parallel host processing and optimizing CPU- and memory-intensive APIs and external service calls.

4x capacity (1,000 to 4,000 hosts), 30% faster

Retry Framework for Distributed Upgrade Workflows

A reusable framework that centralizes transient failure classification and recovery logic for API-triggered operations and asynchronous task polling. Validated with Chaos Mesh pod-failure injection, reproducing customer-observed failure scenarios.

Recovers from downstream outages of up to 12 minutes

Hunting a Production JVM Deadlock

Finding the root cause of a critical deadlock affecting multiple production customers by combining thread-dump analysis, JProfiler profiling, and JMeter load testing to reproduce it.

Permanent fix, zero recurrence

Broadcom (formerly VMware)

Software Engineer · Feb 2022 - Present

Backend engineer on VMware Cloud Foundation, building the services that install, scale, and manage the lifecycle of private cloud infrastructure.

• Owned end-to-end design and delivery of VCF Automation component installation workflows on Kubernetes using Netflix Conductor. Automated cluster bootstrapping and scale-up operations, enabling Day-N component deployments independent of Day-0 provisioning.
• Designed and delivered Fleet Build Service and SDDC Build Service for VCF's greenfield Fleet LCM platform, supporting Day-0 and Day-N installation across 12+ fleet- and instance-level components. Defined and evolved REST API contracts, service boundaries, and persistence models amid changing requirements.
• Extended VCF Brownfield Import to support standalone hosts and single-host clusters, and designed a reusable guardrail framework to validate infrastructure invariants before import and Day-N operations.
• Designed isolated workload domains with independent vSphere SSO domains, increasing workload-domain scalability 67% (15 to 25) and enabling stronger multi-tenant isolation.
• Redesigned topology locking with top-down resource traversal so only required resources are locked, reducing locking overhead by up to 65%.