Low-Level Software Engineer (AI Hardware)

Location: Hybrid, based in Toronto, Canada

Compensation: $100,000–$500,000, including base and variable compensation targets.

Company

Tenstorrent develops AI hardware and software platforms, including custom RISC-V CPUs and high-performance systems for AI workloads.

What you will do

• Design and implement low-level kernels and software primitives for Tenstorrent AI hardware.
• Optimize performance through profiling, bottleneck analysis, implementation, and validation.
• Develop software for efficient model bring-up and execution across Tenstorrent hardware.
• Work with LLK teams across performance, tools, compute, and architecture bring-up.
• Collaborate with hardware and runtime engineers to improve the quality, performance, and capabilities of the low-level software stack.
• Own complex technical problems from design through implementation and optimization.

Requirements

• Strong C/C++ engineering experience building low-level, performance-critical software.
• Deep understanding of computer architecture and how software maps to hardware.
• Strong problem-solving skills for complex performance and systems challenges.
• Ability to work close to hardware and balance performance, correctness, and maintainability.
• Effective collaboration with software and hardware teams and the ability to influence technical decisions.
• Eligibility to access U.S. export-controlled technology is required; employment may depend on citizenship, permanent residency, or prior license approval.

Culture & Benefits

• Work on custom AI hardware, instruction sets, and low-level kernels.
• Learn advanced hardware-aware optimization, performance analysis, and debugging.
• Collaborate across hardware, kernels, runtime, and higher-level software teams.
• Compensation and benefits package with offers adjusted according to experience, skills, education, background, and location.
• Opportunities are available for candidates at various experience levels, with level determined during the interview process.