Performance Architect - AI Hardware
TenstorrentJob Title
Performance Architect - AI Hardware
Role Summary
Join the Tensix team to model, analyze, and optimize real AI workloads on Tenstorrent's compute fabric. You will translate workload behavior into architectural insight, drive performance tradeoffs, and shape hardware features through measurable evaluations.
This role interfaces across architecture, RTL, compiler, and runtime teams to improve efficiency and scalability of next-generation AI systems.
Experience Level
Senior. Candidates at various experience levels will be considered; the interview process will determine the appropriate level and offer.
Responsibilities
The core responsibilities focus on modeling, measurement, and cross-team collaboration to guide hardware design.
- Benchmark and analyze complex AI workloads on single- and multi-node configurations to identify bottlenecks and quantify impact.
- Develop and maintain performance models, simulators, and micro-benchmark suites for feature evaluation.
- Perform detailed PPA (performance, power, area) studies to evaluate design tradeoffs and inform co-design decisions.
- Collaborate with RTL, compiler, and runtime teams to instrument systems and correlate modeling with silicon results.
- Translate deep learning workload characteristics into actionable architectural requirements and metrics.
Requirements
Must-have technical skills and constraints for successful candidates.
- Proficient in C++ and Python for simulation, modeling, and performance analysis.
- Experience building or using performance models, simulators, and benchmarking frameworks for heterogeneous compute systems.
- Experience analyzing multi-node and distributed AI workloads and identifying system-level bottlenecks.
- Ability to perform PPA-style tradeoff analysis and present quantitative recommendations to engineering teams.
- Strong collaboration skills to work across hardware, compiler, and runtime teams and correlate results with silicon.
- Data-driven problem solving and a track record of improving performance or efficiency in complex systems.
- Employment is contingent on eligibility to access U.S. export-controlled technology; offers may depend on citizenship, permanent residency, or ability to obtain required export licenses.
- Nice-to-have: prior experience with AI accelerators, RISC-V systems, or large-scale distributed training inference workloads.
Education Requirements
Not specified.
About the Company
Company: Tenstorrent
Headquarters: Austin, Texas, United States
Tenstorrent is a technology company focused on designing innovative computing solutions. They are known for their expertise in the development of advanced hardware, including ASICs and SoCs, aimed at enhancing performance and efficiency in various applications.
