Head of Performance Visibility

Job Title

Head of Performance Visibility

Role Summary

Lead definition and implementation of performance observability for Etched's ML accelerator platform spanning custom silicon, runtimes, compilers, drivers, and distributed inference environments. This role creates the telemetry, correlation frameworks, and reasoning systems that connect low-level hardware signals to workload- and cluster-level behavior.

This is a hands-on, cross-disciplinary engineering role working with hardware, driver, runtime, and ML infrastructure teams to design scalable instrumentation, time-alignment, counter taxonomies, and analysis tooling.

Experience Level

Senior level. The posting targets an experienced technical leader; explicit years of experience are not specified.

Responsibilities

The Head of Performance Visibility will set architecture and deliver production systems for cross-layer performance observability.

• Define architecture for collecting and structuring telemetry across CPUs, drivers, interconnects, and accelerators.
• Design scalable models for correlating performance events across device and host boundaries.
• Implement time synchronization and trace-alignment strategies across multi-device systems.
• Define structured counter taxonomies and derived performance metrics linking hardware signals to workload behavior.
• Influence instrumentation strategy for future hardware generations.
• Build tools to identify bottlenecks among multi-accelerator workloads within hosts, racks, and clusters.
• Develop cluster-scale performance analysis for distributed inference across data center networks.
• Contribute to analysis engines and developer-facing tooling that transform raw telemetry into actionable insight.
• Lead cross-functional architectural initiatives spanning hardware and software teams.

Requirements

Required technical skills and experience, plus a few preferred areas.

• Must-have: Deep experience building systems at the hardware/software intersection, including hands-on work implementing profiling, tracing, or observability infrastructure (not just product/requirements).
• Must-have: Proven ability to translate raw hardware signals into production-grade telemetry, analysis infrastructure, and performance models.
• Must-have: Experience correlating time-series events and aligning timestamps across distributed systems and multi-device environments.
• Must-have: Systems programming expertise in C++ or Rust with a track record of shipping low-level infrastructure near hardware or runtime systems.
• Must-have: Experience designing distributed correlation mechanisms, timestamp-alignment strategies, and performance modeling frameworks across devices/hosts.
• Must-have: Experience with hardware counter design, instrumentation strategy, and performance modeling for large-scale ML workloads.
• Must-have: Experience with high-performance computing systems or large AI training/inference clusters and leading cross-functional architectural efforts.
• Nice-to-have: Prior work on distributed tracing/observability platforms at scale, familiarity with driver/interconnect architecture, or specific experience in distributed inference networks.

Education Requirements

Not specified.

About the Company

Company: Etched

Headquarters: San Jose, CA, United States

Etched develops purpose-built AI inference ASICs and systems optimized for transformer models, aiming to deliver significantly higher performance, lower cost, and lower latency than GPUs. The company focuses on enabling applications like real-time video generation and advanced reasoning agents, and is backed by leading investors and engineers.