S. Vignesh

S. Vignesh

Auto-generated profile, based on publicly available information

S. Vignesh

Indian Institute of Technology Hyderabad, Hyderabad, India

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2026 in IEEE Transactions on Network and Service Management
Verified
Journal Article
Closed access
SHS. A. Harish
S. A. Harish
SVS. Vignesh
S. Vignesh
DPDivya Pathak
Divya Pathak
ASAnil Kumar Sharma
Anil Kumar Sharma
PTPraveen Tammana
Praveen Tammana

17 September 2026
Verified
Conference Paper
Open access
HAHarish S A
Harish S A
VSVignesh S
Vignesh S
AKAshwin Kurella
Ashwin Kurella
RGRohan Gandhi
Rohan Gandhi
PTPraveen Tammana
Praveen Tammana
While Mixture-of-Experts (MoE) models scale LLM serving across hundreds of GPUs, their reliance on all-to-all communication makes them susceptible to gray failures (e.g., GPU stragglers) which inflate serving latency without explicit errors, complicating fault localization. We present FaultSense, an application-layer approach that localizes faulty GPUs and communication paths in MoE model serving replicas without host instrumentation. FaultSense introduces a lightweight MoE probe model prototype alongside a two-phase hierarchical fault localization algorithm. Modeling the cluster as a GPU communication graph, it performs: (1) a global coverage that tests all GPUs and communication paths with a small set of probes to identify suspicious components, and then (2) a drill-down into suspicious components to localize faulty GPUs or communication paths. Our design reduces the number of diagnostic tests compared to individually testing all components (up to 20×) while maintaining low memory overheads.