Etched
MTS, Inference Performance Visibility
$150K–$400K
San Jose · On-site · Full-time · Posted Jan 18, 2026
About the job
Design and build performance analysis tools for Etched’s custom ML accelerator, helping engineers and customers understand workloads and identify hardware bottlenecks.
What you'll do
- Architect performance analysis tooling, including data collection, processing, analysis, and CLI or GUI interfaces.
- Capture and correlate hardware, host, driver, PCIe, and system performance data; analyze it to identify bottlenecks.
- Develop visualizations and collaborate across hardware, firmware, driver, compiler, and ML application teams.
What they ask for
- Strong proficiency in C++ or Rust.
- Deep understanding of computer architecture, memory hierarchies, and interconnects, especially PCIe.
- Proven experience with low-level performance analysis, profiling, and bottleneck identification on complex hardware systems.
- Experience with performance analysis tools and working close to hardware, such as reading performance counters or interacting with device drivers.
Nice to have
- Direct experience developing performance analysis or debugging tools.
- Experience with ML accelerator architectures, kernel-mode drivers, compiler internals, PCIe analysis, multi-chip or multi-host systems, firmware, or hardware description languages.
Pay
$150K – $400K (base salary)
Tools
C++, Rust, Python, NVIDIA Nsight, AMD uProf, Intel VTune, perf, Tracy
AI focus
Build performance analysis tooling for an ML accelerator used for inference workloads.
Model-written summary · the original posting is authoritative