Principal Engineer with 11+ years of experience in distributed storage systems and performance engineering, specializing in IO Datapath analysis, cross-layer performance debugging, and large-scale workload characterization across the VMware vSAN storage stack.
Experience
2024 — Now
2024 — Now
Bellevue, WA
* Led investigation of IO performance degradation on modern multi-die CPU architectures - like Intel Granite Rapids (GNR), Intel Emerald Rapids (EMR) and AMD Genoa and Turin. Insights helped leadership make decisions about next-generation platform optimization and deployment.
-Root caused data locality loss due to suboptimal scheduling of related processes placed across CPU sockets.
-Micro-benchmarking and application-level performance experiments identified critical processes to be co-located.
• Analysis guided NUMA-aware scheduling improvements and validated performance gains in real workloads.
-Developed a cross-platform visualization tool to map world placement across NUMA nodes on Intel and AMD systems, enabling engineers to quickly diagnose data locality issues across diverse hardware environments.
* Drove end-to-end IO Datapath performance validation for integrated VMware Cloud Foundation (VCF) deployments, to ensure production-scale performance readiness.
-Evaluated vSAN performance over Smart NIC-accelerated networking (NSX-ENS), analyzing effects of Large Receive Offload (LRO) and VXLAN offloads on storage IO latency and throughput.
-Debugged and root-caused cross-layer IO performance regressions across storage, networking, and NIC drivers, isolating bottlenecks affecting integrated platform deployments.
-Served as technical liaison between storage and networking teams, resolving cross-stack issues and strengthening integration for future VCF releases.
* Modeled customer production workloads using benchmarking tools to reproduce performance issues and validate optimizations in controlled environments
* Tuned application and storage benchmarks (HammerDB) to characterize system behavior under different workload patterns.
2018 — 2024
2018 — 2024
Palo Alto, CA
VMware Cloud deployments on AWS Performance (2020 – 2024)
* Team-lead for engineers responsible for end-to-end storage performance qualification of VMware Cloud deployments on AWS (VMC)
- Performed benchmarking and scalability analysis on i3.metal, i3en.metal, and i4i.metal NVME based EC2 instances.
- Performed large-scale performance experiments and regression validation
- Upgraded and stabilized automation harnesses for performance qualification to expedite VMC release cycle.
* Delivered performance readiness reviews and deep-dive analysis to PMs, performance architects, and executive leadership.
* Designed targeted workloads to surface infrastructure bottlenecks and isolate root causes across storage and networking layers. Identified key issues like disk latency degradation and network-level corruption on AWS.
Distributed Storage Software Development (2018 – 2020)
* Worked on the distributed storage stack, primarily in the LFS layer, focusing on feature development, debugging, and bug fixing
* Implemented and maintained product features including
Improved log-structured layer stability by redesigning back pressure mechanism (Log smoothening), adaptive delete IO throttling
* Optimized a critical vSAN IO Datapath by increased batching and parallelization, improving POC workload performance by ~30% and enabling broader vSAN adoption.
* Diagnosed and resolved a complex memory leak in production, delivering detailed root-cause analysis (RCA) to enterprise customers, including in-person briefings for high-profile accounts
* Recognized with customer Success award for my above contributions
* Contributed to and enhanced the Grafana + Influx DB performance observability platform, enabling real-time visualization and analysis of distributed storage performance metrics.
2015 — 2018
2015 — 2018
Palo Alto, California, United States
* Co-developed trace Analyzer, a tracing-based diagnostic tool that stitches IO traces across hosts and multiple layers of the storage stack to graphically identify latency outliers and bottleneck layers.
* Developed GDB macros and enhanced internal testing frameworks, cutting testing time by ~40% and accelerating feature shipping for the team.
* Provided training and triaging playbooks for Customer support teams for vSAN performance.
2014 — 2015
2014 — 2015
Raleigh-Durham, North Carolina Area
Worked as QA engineer for the SAN functional team and was closely associated towards understanding, planning and testing for different feature releases of ONTAP specific to SAN. Developed automation for feature testing and stressing internal components for integrity and consistency. ( Role : Perl Developer)
2013 — 2013
2013 — 2013
Raleigh
Rally-BURT connector ( Role : Perl Developer)
Worked towards developing an in-house tool for syncing changes between 2 applications which were referenced differently by different groups within the company. The idea was to provide increased collaboration and communication between mutually dependent yet exclusive groups.
Education
University of California, Berkeley, Haas School of Business
Product Management Certificate Program
North Carolina State University
Master's degree
YMCA Institute of Engineering