Network Site Reliability Engineer, AI Solutions Lab - Santa Clara, CA, United States
$131,010 - $196,300 USD yearly
Originally posted 16 October 2026 by the employer.
This role owns the compute and software foundation for validating Marvell solutions across customer-relevant workloads.
About the role
This role is responsible for building, standardizing, and operating environments to validate Marvell solutions across customer-relevant workloads, including scale-up, scale-out, memory acceleration, and DPU-enabled use cases. The team operates GPU/HPC lab clusters and validates Marvell's Teralynx switching and AI fabrics.
What you'll do
- Own compute provisioning, cluster bring-up, OS images, ROCm and NIC driver stacks, and environment standardization for lab systems.
- Build and maintain MLPerf-style harnesses, workload libraries, test plans, result pipelines, and dashboards to support repeatable validation.
- Establish baseline observability for cluster health, job execution, failures, utilization, and regression tracking.
- Deploy and manage workload scheduling and orchestration (Kubernetes and/or Slurm) for running validation and benchmark jobs across the lab.
- Own incident response for lab infrastructure, driving root-cause analysis and reliability improvements after cluster or job failures.
- Improve repeatability and operational efficiency through automation, infrastructure, code, and standardized workflows.
What you'll need
- MS in Electrical/Computer Engineering, Computer Science, or related field; 6+ years of experience as a software/systems engineer, with at least 4+ years of Linux/infrastructure experience.
- Strong Linux systems experience at scale, including solid data center and cluster operations fundamentals.
- Proficient in Python/TCL scripting and automation, with familiarity in C/C++ API development; proficiency in Python specifically for metrics, observability, and pipeline automation.
- Experience with automation tools such as Ansible and Terraform.
- Experience building and maintaining test harnesses, workload libraries, test plans, result pipelines, and dashboards to support repeatable validation.
- Hands-on experience with workload scheduling and orchestration (Kubernetes and/or Slurm) in a lab or production environment.
Nice to have
- Experience operating AI or HPC clusters in a lab or production environment.
- Working familiarity with a distributed training framework (PyTorch DDP/FSDP, NeMo, or similar).
- Working knowledge of at least one distributed-storage or parallel filesystem technology (e.g., Lustre, GPFS, BeeGFS, Ceph).
- Basic experience with model tuning and workload optimization.
Skills: compute provisioning, cluster bring-up, OS images, Kubernetes, Linux systems
This role has been open 0 days — well below the 51-day median for Infrastructure/Platform Software roles.
Infrastructure/Platform Software · Infrastructure Platform
|
Open roles in category
974
|
Median days open
51 d
|
Median salary
$236k
|
See the full market breakdown ▾Category comparison, skills in demand, and who else is hiring
| Metric | Marvell Technology | All employers we track in this specialty (974 roles · 86 employers) |
|---|---|---|
| Open roles in this specialty | 10 | 974 |
| Open roles in the wider Software, Firmware & Systems family | 72 | 6157 · 145 employers |
| Median days open | 4 d | 51 d (−47 d vs this employer) |
| Median salary (USD postings) | — | $236k |
Skills observed across this category: compute provisioning, cluster bring-up, OS images, Kubernetes, Linux systems
Who's hiring in this category
- NVIDIA · 281 open roles · median 46 d
- Qualcomm · 61 open roles · median 55 d
- AMD · 55 open roles · median 22 d
- Graphcore · 44 open roles · median 91 d
- Intel Corporation · 41 open roles · median 10 d
- Cerebras · 39 open roles · median 81 d
How we counted: 974 open Infrastructure/Platform Software (Infrastructure Platform) roles from 86 employers tracked in the SemiconductorJobs index, counted 16 Oct 2026. Specialty figures count only roles carrying this exact specialty label, so an employer's related work in neighbouring specialties is not included there — it is counted in the wider Software, Firmware & Systems family row. Figures refresh nightly.