Drimva

NVIDIA

Principal Software Engineer, At-Scale Reliability and Fleet Intelligence — CSP Engagements

$272K–$431K

US, CA, Santa Clara · Principal · Full-time · Posted Oct 13, 2026

Experience
15+ years in systems software at datacenter scale, or reliability engineering focused on at-scale challenges.
Education
BS or MS in Computer Science, Electrical Engineering, Statistics, or related field (or equivalent experience)

About the job

Lead fleet-scale reliability work with hyperscale customers, analyze reliability data, and drive improvements to NVIDIA platforms and monitoring systems.

What you'll do

  • Coordinate reliability work streams with CSP teams, including MTBI methodology, health monitoring, and launch readiness.
  • Analyze fleet failure patterns and telemetry, then champion improvements across firmware, driver, hardware, and quality teams.
  • Define burn-in and certification criteria and develop predictive failure models validated in customer environments.

What they ask for

  • Deep expertise in multi-NUMA, rack-scale system software and firmware, and statistical failure analysis.
  • Experience with fleet-level telemetry and observability systems, including time-series databases, anomaly detection, health scoring, or event correlation.
  • Understanding of hardware failure modes in large-scale GPU or accelerator deployments and experience with burn-in, stress testing, or certification frameworks.
  • Strong communication and demonstrated success driving cross-functional improvements across hardware, firmware, and software teams.

Nice to have

  • Fleet reliability experience at a hyperscaler; familiarity with NVIDIA GPU error taxonomy.
  • Experience building health-scoring or predictive failure models for accelerator or HPC infrastructure.

Pay

272,000 USD - 431,250 USD (base salary)

Tools

MTBI, MTBF, Pareto, survival analysis, Weibull, time-series databases, Xid errors, NVLink error counters

AI focus

Develop predictive failure models using fleet telemetry and validate their effectiveness in customer environments.

Model-written summary · the original posting is authoritative

First seen Oct 15, 2026 · Last verified Nov 5, 2026 (UTC)