Performance Engineer - San Jose, California, United States
$135,000 - $170,000 USD yearly
Originally posted 24 July 2026 by the employer — open 48 days.
About the role
This role involves defining performance measurements for rack-scale AI infrastructure.
What you'll do
- Establish theoretical and measured roofline models for scale-up fabric.
- Build and maintain baseline performance benchmarks using NVBandwidth and NCCL.
- Quantify the impact of differentiated AI fabric features against baselines using synthetic benchmarks and real inference workloads.
- Run end-to-end inference model workloads on target hardware.
- Evaluate fabric performance as inference cluster size scales from 16 to 32 GPUs and beyond.
- Design and execute head-to-head performance comparisons against competing fabric switch solutions.
- Design, build, and maintain automated lab infrastructure including test execution pipelines, traffic generation tooling, and data collection and reporting systems.
- Enable repeatable, high-quality, and scalable performance measurements across all hardware configurations.
What you'll need
- 8+ years of experience with performance analysis and optimization of computing and networking systems.
- Expertise in GPU system architecture and fabric interconnects like PCIe and Ethernet.
- Experience with standard benchmarks (e.g., MLPerf, ResNet-50) and low-level performance tools (e.g., NVBandwidth, NCCL).
- Proficiency in Python and C++.
- Experience with data analysis and visualization tools.
- Deep understanding of high-speed interconnect protocols such as PCIe, CXL, and Ethernet.
- Experience designing, developing, and deploying performance measurement automation in a lab environment.
Skills: GPU clusters, AI/ML workloads, scale-up fabric, PCIe, Ethernet
This role has been open 47 days — well below the 121-day median for Hardware Systems Engineering roles.
Hardware Systems Engineering · Serdes Highspeed
|
Open roles in category
91
|
Median days open
121 d
|
Median salary
$197k
|
See the full market breakdown ▾Category comparison, skills in demand, and who else is hiring
| Metric | Astera Labs | All employers we track in this specialty (91 roles · 33 employers) |
|---|---|---|
| Open roles in this specialty | 2 | 91 |
| Open roles in the wider Software, Firmware & Systems family | 18 | 5650 · 142 employers |
| Median days open | 27 d | 121 d (−94 d vs this employer) |
| Median salary (USD postings) | — | $197k |
Skills observed across this category: GPU clusters, AI/ML workloads, scale-up fabric, PCIe, Ethernet
Who's hiring in this category
- NVIDIA · 14 open roles · median 154 d
- Qualcomm · 12 open roles · median 154 d
- AMD · 9 open roles · median 77 d
- Keysight Technologies · 6 open roles · median 107 d
- Marvell Technology · 6 open roles · median 164 d
- Etched · 4 open roles · median 156 d
How we counted: 91 open Hardware Systems Engineering (Serdes Highspeed) roles from 33 employers tracked in the SemiconductorJobs index, counted 10 Sept 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.